Red Hat discloses RHEL roadmap (TechTarget) [LWN.net]

TechTarget has an interview with Denise Dumas, Red Hat's director of software engineering, about RHEL 6.5 and 7. In it, Dumas outlines some changes coming in those releases, particularly in the areas of storage, networking, in-place upgrades from RHEL 6, and the default desktop:

We think that people who are accustomed to Gnome 2 will use classic mode until they're ready to experiment with modern mode. Classic mode is going to be the default for RHEL 7, and we're in the final stages now. We're tweaking it and having people experiment with it. The last thing we want to do is disrupt our customers' workflows.

I think it's been hard for the Gnome guys, because they really, really love modern mode, because that's where their hearts are. But they've done a great job putting together classic mode for us, and I think it's going to keep people working on RHEL 5, 6 and 7 who don't want to retrain their fingers each time they switch operating systems -- I think classic mode's going to be really helpful for them.

to post comments

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 13, 2013 21:32 UTC (Thu) by cyperpunks (subscriber, #39406) [Link] (135 responses)

Finally enterprise had a vote in the GNOME fiasco...

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 2:37 UTC (Fri) by bojan (subscriber, #14302) [Link] (128 responses)

Chuckle... The Gnome 3 "overview" is such a great invention that Red Hat (company behind Gnome) decided to leave it out of the default setup in their money-spinner. Yep, that is what I call a vote of confidence. :-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 7:04 UTC (Fri) by alexl (guest, #19068) [Link] (1 responses)

Classic mode has the overview

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 7:08 UTC (Fri) by bojan (subscriber, #14302) [Link]

Yep, hidden away behind what looks like Gnome 2-- (which essentially means you never have to use it).

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 7:18 UTC (Fri) by dsommers (subscriber, #55274) [Link] (125 responses)

> The Gnome 3 "overview" is such a great invention that Red Hat (company
> behind Gnome)

Red Hat initiates a lot of good projects, I'll grant you that. But Red Hat is probably not the company behind GNOME. They contribute a lot to GNOME though, as well as many other projects. The first Red Hat based distribution with GNOME preview was Red Hat Linux 5.1 (not Red Hat Enterprise Linux) [1] which was released May 22, 1998 [2].

The GNOME Project was started in 1997 by two then university students, Miguel de Icaza and Federico Mena [3]. In 1999, de Icaza, along with Nat Friedman, co-founded Helix Code, a GNOME-oriented free software company that employed a large number of other GNOME hackers. In 2001, Helix Code, later renamed Ximian ... In August 2003, Ximian was acquired by Novell. [4]

Just to get the history straight ...

[1] <http://linuxgazette.net/165/laycock.html>
[2] <https://en.wikipedia.org/wiki/Red_Hat_Linux>
[3] <https://www.gnome.org/about/>
[4] <https://en.wikipedia.org/wiki/Miguel_de_Icaza>

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 7:41 UTC (Fri) by bojan (subscriber, #14302) [Link] (104 responses)

Just what difference does it make in 2013 who started the project in 1997?

My point being this: Red Hat _pays_ folks to write Gnome. It is their default desktop environment.

PS. Just because other companies do that too does not mean Red Hat is not behind it.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 8:46 UTC (Fri) by dsommers (subscriber, #55274) [Link] (103 responses)

Red Hat is surely and clearly involved in the GNOME project, and has been for quite some time. But your statement is easily understood that Red Hat *controls* the project. I would say that's a fairly unbiased statement.

In 2010 this GNOME census [1] presentation was given. And it surely puts Red Hat high up in the contributors list, but it is actually *not* the biggest player in this game all in all. Red Hat is given credit for 16.30% of the commits, while there are 16.94% of unidentifiable companies (I presume) and 23.45% of the commits were from VOLUNTEERS. These three groups sums up to ~57% of all commits to the GNOME project. Red Hat's contribution in this context is a little bit less than 1/3 of the top 3. And then you have ~43% of commits from contributors which is not among the top 3.

So to say that Red Hat is behind GNOME is fairly unfair to the majority of contributors. I'm sure Red Hat is proud of their GNOME contributions, but they probably don't want (or need) to take the glory from the other contributors as well.

[1] <http://blogs.gnome.org/bolsh/2010/07/28/gnome-census/>

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 9:11 UTC (Fri) by nim-nim (subscriber, #34454) [Link] (10 responses)

While Red Hat does not control the project, its minority stake is clearly big enough to have some form of veto power.

16,30 % of commits is huge and it undercounts all the work done by third parties because of their stake in the RHEL/Centos/Fedora ecosystem, all the work Red Hat employees declare as "volunteer" (but would they still be as interested in GNOME if they were not involved with it at work? when Nokia switched from GTK to QT lots of people in the Nokia ecosystem discovered they weren't that interested in GNOME anymore), all the synergies between Red Hat GNOME committers and Red Hat people working on other layers of the software stack.

Besides, the second big corporate partner is Suse (10,44%), but Suse customers ask it to diverge as little as possible from Red Hat core choices (they want the ability so switch supplier without retraining), any Red Hat divestment would be followed by a similar move Suse-side.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 10:03 UTC (Fri) by ovitters (guest, #27950) [Link] (9 responses)

Could you point out where Red Hat has veto power? Maintainers have influence, not companies. If companies hire maintainers they can influence, not by being a company.

That a maintainer can set direction for their module is pretty common.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 11:21 UTC (Fri) by nim-nim (subscriber, #34454) [Link] (8 responses)

When you have enough influence, you have veto power. Though a project always has the possibility to ignore inconvenient stakeholders, and that typically does not end well (see xfree86, apache openoffice, etc).

The commit stats clearly show Enterprise Linux distributions have a key stake in GNOME. Together Red Hat and Suse pay directly for more than 25% of the GNOME work (not counting the work of affiliates). No other entity is ready to pay enough people to work on it, to manage more than 5% of commits.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 11:32 UTC (Fri) by ovitters (guest, #27950) [Link] (7 responses)

I asked to show where Red Hat has veto power. You assert this "vero power" again without showing any references.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 12:09 UTC (Fri) by nim-nim (subscriber, #34454) [Link] (6 responses)

I replied with facts. You chose to ignore them. What's the point of your message exactly?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 12:27 UTC (Fri) by ovitters (guest, #27950) [Link] (3 responses)

I've asked various times to show proof of your assertion that Red Hat has "veto power" in the GNOME project. You show all kinds of things, but nothing which backs up your claim.

To be clear: The entity having veto power is GNOME board and GNOME release team. I am a member of the release team. I am pretty aware of the rules and how things are done.

I have never seen anything about "veto power". I've asked you to back this up various times, you haven't shown anything.

The only point you could make is that they've hired maintainers, as I said before. However, that is not "veto power".

So please back up your claims.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 14:23 UTC (Sat) by sbergman27 (guest, #10767) [Link] (2 responses)

Red Hat can veto by laying down the law and choosing not to use your version of Gnome in RHEL. Red Hat made it clear to you that Gnome3 shell was unacceptable, and you jumped. The recent course of events is, itself, proof of nim-nim's assertion. Red Hat is a billion dollar company, and not a hobby project. When they invest in a project like Gnome, they reasonably insist upon effective control. This does not necessarily mean that the control will be explicitly written into the bylaws. In fact, it's often best that the veto power *not* make it into the bylaws. But the power is, effectively, there. And that is no accident.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 16, 2013 22:41 UTC (Sun) by dag- (guest, #30207) [Link] (1 responses)

So you are basically saying that Red Hat has veto powers over what they put into RHEL. And that affects also how they customize the Gnome that goes into RHEL.

Makes sense, since that's what they have to support for the next 13 years (starting from GA).

But that doesn't mean Red Hat has veto powers over Gnome (either the project, or the source code). They may have influence, but influence is still not veto power...

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 16, 2013 22:53 UTC (Sun) by sbergman27 (guest, #10767) [Link]

10 years ago, I might have agreed that what the Gnome guys decided was more important than what was decided for them by higher ups.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 12:38 UTC (Fri) by AndreE (guest, #60148) [Link] (1 responses)

What facts? You state a direct link between number of commits and right to veto features, but offer no proof for such a link. Is there an actual example of a GNOME roadmap or release being explicitly dictated by the needs of Red Hat?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 18:45 UTC (Wed) by jwarnica (subscriber, #27492) [Link]

He isn't saying that there is some coded rule or bylaw, just observing that if 16% of the contributors refuse to do something, then its very very unlikely its going to happen.

"When you do things right, people won't be sure you've done anything at all."

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 9:22 UTC (Fri) by bojan (subscriber, #14302) [Link] (91 responses)

And where exactly did I say that it was _only_ Red Hat that was behind Gnome? I just said they were behind it. You numbers prove it rather neatly, actually.

Red Hat shipping Gnome in classic mode in RHEL7 is roughly the equivalent of Microsoft shipping Windows 8 in desktop mode, after spending all the years developing the tiles.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 12:56 UTC (Fri) by dsommers (subscriber, #55274) [Link] (90 responses)

> And where exactly did I say that it was _only_ Red Hat that was behind 
> Gnome? I just said they were behind it. You numbers prove it rather 
> neatly, actually.

It's very easy to do that interpretation from your first statement:

    The Gnome 3 "overview" is such a great invention that
    Red Hat (company behind Gnome) decided to leave it out
    of the default setup in their money-spinner. Yep, that
    is what I call a vote of confidence.

Here it sounds like you claim Red Hat carries the responsibility for GNOME 3, and even doesn't have the courage to bring the full GNOME 3 experience to their customers.

And that is what I reacted to. GNOME is a project, where Red Hat is a participant, pretty much on the same level as everyone else who wants to contribute. That Red Hat decides to not enable a new feature by default for their Enterprise customers, doesn't mean they don't have faith in it. But they obviously listens to their customers' concerns, and try to avoid upsetting them. And for customers who want that new experience, they can enable it when they are ready for it. That's something completely different than claiming Red Hat doesn't have faith in something.

If Red Hat hadn't had faith in the GNOME 3, why would they ship it at all? Why would they keep on contributing to the GNOME project?
(those questions are rhetorical questions)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 3:32 UTC (Sat) by bojan (subscriber, #14302) [Link] (70 responses)

> That Red Hat decides to not enable a new feature by default for their Enterprise customers, doesn't mean they don't have faith in it.

Now you are being hilarious. The whole paradigm of Gnome 3 is the overview thing. The new "distraction free philosophy" of the desktop or some such nonsense. It is considered so valuable by Red Hat that they decided to ship a cut down Gnome 2 look instead. Yeah.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 3:51 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (69 responses)

"The whole paradigm of Gnome 3 is the overview thing"

Says who? None of the design documents from GNOME project focus on this as a primary aspect.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 4:10 UTC (Sat) by bojan (subscriber, #14302) [Link] (31 responses)

I see the 1984 mode is well and truly on. History being rewritten:

https://live.gnome.org/GnomeShell/Design#Activities_Overview
https://live.gnome.org/GnomeShell/Design/FAQ#Why_no_windo...

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 4:28 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (30 responses)

Even the first link you provided shows that overview isn't the "whole paradigm. It is just one of the design elements. Other prominent ones include dash, notifications etc.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 4:55 UTC (Sat) by bojan (subscriber, #14302) [Link] (29 responses)

Ah, yes. Another rewrite of history. The "distraction free" philosophy never happened. We all just imagined that:

https://live.gnome.org/GnomeShell/Design#Goals_and_advant...

The paradigm of Gnome 3 (which is to provide distraction free environment, whatever is that supposed to mean) absolutely depends on the existence of overview (in the mind of Gnome developers). In fact, dash is an artefact in overview only. Notification are just another type of overview (as you cannot see them in normal view at all once they magically disappear).

Coming back to the original point. Amazing efforts have been expended to provide this supposed distraction free environment, only to default to distraction abundant environment in RHEL7 (taskbar, applications, places, poor replacement for workspace switcher etc.) to poor users. I'm sure their heads are going to explode now, given they won't be distraction free. :-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 5:00 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (9 responses)

You are repeatedly drawing up a strawman. I suggest you read what I am replying to you rather than assume things I don't say just to argue against your own assumptions. I am asking you whether you can substantiate that overview is the primary paradigm of GNOME Shell as opposed to one of the design elements and it appears you cannot.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 5:40 UTC (Sat) by bojan (subscriber, #14302) [Link] (8 responses)

I cannot help you if you cannot see this from docs and discussions that are publicly available. Also, I cannot help you if you cannot see that bringing back all the "bad" elements by default is essentially saying that the whole thing (overview, which enables "distracton free") was a mistake.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 6:56 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (4 responses)

http://extensions.gnome.org is one of the best parts of GNOME 3 and I would much less productive without it.

GNOME project provided classic mode (which itself is just a bunch of extensions) to meet the needs of users who prefer the traditional UI elements and this matches the nature of RHEL. It demonstrates the flexibility of GNOME Shell and power of extensions to virtually modify any UI element and they are much more easier to develop compared to the panel applets in GNOME 2.x. The proof is in the sheer number of extensions that do things GNOME 2.x never could and I use several of them.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 13:18 UTC (Mon) by nix (subscriber, #2304) [Link] (3 responses)

> http://extensions.gnome.org is one of the best parts of GNOME 3

Curious. A whole lot of GNOME devs back when extensions were introduced were saying they were against them, that they 'diluted the GNOME brand', that you should never rely on them and that they'd get broken as often as possible, and that they'd try to get them removed as soon as possible.

And now all of a sudden it's 'one of the best parts of GNOME 3'. I see some divergence of opinion here...

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 14:55 UTC (Mon) by raven667 (guest, #5198) [Link] (2 responses)

I'm sure there are some GNOME devs who feel that way but that clearly isn't a majority or dominant opinion otherwise extensions wouldn't be hosted on gnome.org and Classic UI wouldn't be part of upstream.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 15:38 UTC (Mon) by andresfreund (subscriber, #69562) [Link] (1 responses)

Well, the designers of gnome 3 were pretty much against it (https://mail.gnome.org/archives/gnome-shell-list/2011-Jun...), I think that does say something about their ideals. That they were overruled is great, but that that was necessary isn't good.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 23:16 UTC (Wed) by sramkrishna (subscriber, #72628) [Link]

There was a notion that the default look of GNOME was something that should be distinguishable. I think we've sort of evolved from that. Extensions is a great 'get out of jail free' card'. It'll still be some time before we see some interesting extensions as to the myriad of extensions that implement a missing feature or add trivial ones.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 9:06 UTC (Sat) by alankila (guest, #47141) [Link] (2 responses)

For whatever little it is worth, I really do prefer the distraction-free approach. Call it the tablet/phone effect: I suddenly discovered that windows suck and in general try to only run fullscreen applications these days. The experience is best with OS X, but GNOME 3 comes a very close second. My main gripe is that not all applications have integrated with the top toolbar (which can be configured to vanish away, leaving *all* screen space for applications).

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 9:16 UTC (Sat) by bojan (subscriber, #14302) [Link]

Lucky you. For the rest of us that have to see multiple windows at the same time (sometimes 8 or 9), this simply does nothing. Making a UI that does single tasking well is probably the easiest problem to solve. It has been done to death in the 90s.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 23:29 UTC (Sat) by mathstuf (subscriber, #69389) [Link]

I've found something similar. I used to have rules to tell KWin to hammer sizes, locations, etc. of windows to different workspaces. Now I use a much more fluid setup with XMonad which gets me the "full screen" by default, but also handles the "need a couple of things at once" scenario automatically. If you really want 100% of the screen for applications, you also need to get rid of those pesky window decorations ;) .

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 5:13 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (18 responses)

To answer your other incorrect assertion, distraction free primarily refers notifications and notifications don't require you to access the overview. You just move the cursor to the bottom (pressure sensitive) or press system + m key (acts as a toggle in GNOME 3.8). Also I use dash as a dock and don't access the overview for that. I rarely use overview at all. I never have heard any GNOME designers treat it as the primary design element and I even attended one of the earliest talks about it in person.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 5:44 UTC (Sat) by bojan (subscriber, #14302) [Link] (17 responses)

No assertion, docs. Also, an extension that gives you the opposite of design goals (read about the evil taskbar), proves my point, not yours.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 5:56 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (16 responses)

By your own admission, you don't really understand why they call it distraction free anyway and it makes no logical sense at all why one would call overview as distraction free. Despite your claims, the docs don't show any connection between the two. Just so that you understand it better, design elements like the black chrome, symbolic icons and notifications are meant to minimize the focus on the chrome and draw attention to the application themselves. That is why it is called distraction free. From a personal perspective, while I like the symbolic icons, I think the notifications system needs some improvements.

The fact that GNOME Shell was designed to provide the flexibility to use a GNOME Shell extension hosted by GNOME project itself proves your point that overview is the primary paradigm? Also, the dash to dock just lets the user access the dash directly without the overview for users who prefer that model. It doesn't change the nature of the dash into a full blown panel or "taskbar".

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 8:55 UTC (Sat) by bojan (subscriber, #14302) [Link] (15 responses)

Direct quote from shell design docs:

"The separation of the overview from the normal window view is a reflection of users' natural focus-switching behaviour. It aims to ensure that users are not distracted when they are occupied with a task and to give them quick access to a streamlined focus-switching interface when they need one. A key feature of the overview is that it allows a user to optionally appraise their current activities prior to making a decision on where to turn their focus to next."

Look, we can do this back and forth all day. Few facts:

- Gnome 3 stated some "design goals" (i.e. philosophy)
- classic mode confirmed this cannot always be followed
- Red Hat confirmed they will use classic mode by default

Conclusion: practical usability is far more important than forcing users' hand into nebulous philosophical claims. I agree.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 9:12 UTC (Sat) by bojan (subscriber, #14302) [Link]

Sorry, head, not hand.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 16, 2013 21:47 UTC (Sun) by rahulsundaram (subscriber, #21946) [Link] (13 responses)

I am glad to hear that you agree with your own conclusions. Red Hat however just stated that they are shipping classic mode by default for continuity for enterprise customers. If they didn't think the default mode for GNOME Shell was useful, they wouldn't be funding it.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 6:51 UTC (Mon) by bojan (subscriber, #14302) [Link] (12 responses)

Wow! So, the new UI is so good that the paying customers will be enjoying the benefits of the brokenness (according to shell design docs) of the emulation of old UI. Yeah, that makes sense.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 7:26 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link] (11 responses)

I guess you have no idea why Red Hat is funding Gnome Shell. Keep trying

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 0:08 UTC (Tue) by bojan (subscriber, #14302) [Link] (10 responses)

Whatever the original reason was, they just said (money talks) that the overview UI isn't it.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 0:12 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link] (9 responses)

There is cognitive dissonance at play here when you ignore the part where the same money is funding GNOME Shell default mode as well.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 0:34 UTC (Tue) by sbergman27 (guest, #10767) [Link]

Red Hat made it clear to the devs that "modern" (LOL!) was not anywhere near usable enough to make it into RHEL7, after even so many years of development. They laid down the law. Red Hat has a responsibility to their customers to shield them from their Fedora children.

But the developers are still valuable enough to retain. Red Hat would prefer them not to leave in a snit.

You can try to argue against that. But isn't this all pretty obvious?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 0:37 UTC (Tue) by bojan (subscriber, #14302) [Link] (7 responses)

No cognitive dissonance at all. They paid some folks money to go do things. These folks came back with a result. Red Hat said, sorry - our users are not going to be subjected to this (overview UI).

In the real world, occasionally when you fund something, you get a lemon. It happens. So, when you get handed that lemon, what do you do? You make lemonade (read: Gnome Classic).

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 0:43 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link] (6 responses)

If your assumptions are correct, they should only fund classic mode going forward. I doubt that.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 0:58 UTC (Tue) by bojan (subscriber, #14302) [Link] (5 responses)

What a completely disingenuous comment.

If some random YouTube reviewer that didn't understand what classic mode was wrote that, I would say fine - the guy doesn't know any better. But you know better, because, as you pointed out out on these very pages, classic mode is just a bunch of extensions thrown together, so that the UI looks a bit like Gnome 2 (and this is what we are talking about here - the UI paradigm shipped by default in RHEL7).

In terms of the platform development, the horse has bolted. Gnome 3 is the new platform (the one Red Hat are behind anyway), so they have to work with what they have (i.e. paid to be built). Saying "only fund classic mode going forward" is a complete nonsense statement for an openly and dedicatedly open source company like Red Hat. And you know it.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 1:05 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link] (4 responses)

Aha, now you begin to see my point. Red Hat is a dedicated open source company and it is foolish to say money talks and only commercial enterprise customers matter. Commercial validation is important but community is relevant to the conversation as well.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 1:15 UTC (Tue) by bojan (subscriber, #14302) [Link] (3 responses)

Your point is that what Red Hat decided to ship as a default UI in RHEL7 is not a reflection on the value of the new, overview based UI. I completely disagree with that point. It is a value statement. If they saw great value in this new paradigm (which they helped finance), they would be happy to shout from the rooftops that they "nailed it". Hey, Microsoft sure did that with tiles.

Red Hat will continue to finance open source because they think they can get what they want cheaper that way. This, however, does not mean that everything they get a as a result will be to their liking. Or that they will risk exposing their _paying_ customers to it. This is where money talks.

Of course, you know all this. You are just trying to defend you position with disingenuous statements now.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 1:18 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link] (2 responses)

"Your point is that what Red Hat decided to ship as a default UI in RHEL7 is not a reflection on the value of the new, overview based UI"

Not all all. I explicitly said commercial validation is important but the focus on only that is too narrow and reeks of proprietary vendors and you need to understand and ack the community value as well.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 1:44 UTC (Tue) by bojan (subscriber, #14302) [Link] (1 responses)

What you actually wrote is this (i.e. exactly what I paraphrased above):

> If they didn't think the default mode for GNOME Shell was useful, they wouldn't be funding it.

Clearly, based on a alternative default UI choice in RHEL7, default mode (overview paradigm) is not what they find useful. Otherwise, they would be promoting as the "best since sliced bread, what Microsoft did with tiles, kinda thingy".

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 1:49 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link]

I understand your argument but find it a sad and narrow perspective. I am just reminded of Solaris and CDE. Hopefully the Linux community on the whole has a broader view of things.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 10:17 UTC (Sat) by hadrons123 (guest, #72126) [Link] (36 responses)

@rahul They(gnome) don't focus overview as the primary aspect. But that's the main thing that everyone(users) is irritated about gnome-shell.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 16:46 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (35 responses)

Everyone? That is categorically false statement. Let's not go overboard.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 17:06 UTC (Sat) by sbergman27 (guest, #10767) [Link] (30 responses)

It's certainly a notably popular Gnome3 show-stopper. Hadrons123 makes a good point.

How ironic. We've had this conversation before... about KDE4. But you were defending KDE, and saying that Gnome2 had been just as disruptive. I maintained that Gnome2 was always reasonably well done, if slightly spartan, at first, but that KDE4 was just terrible. Gnome3-shell is worse than anything which has ever come before. Gnome2 is as polished as ever. I haven't kept up with KDE4. But if it came down to Gnome3-shell or KDE4 for my users, and KDE4 didn't pan out for some reason... then.. well... there's always FVWM2 and AnotherLevel. (And no, I can't believe I said that. ;-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 17:11 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (29 responses)

Good that you noted that I have argued for KDE 4 and just as I predicted, KDE 4 has settled down and users have accepted it. I distinctly remember how disruptive that GNOME 1 to GNOME 2 transition was for many users till several revisions later. Many users deflected then and among those who stayed some apparently have become such huge GNOME 2 fans that GNOME 3 UI model is a problem for them. 3.8 apparently has convinced some users that GNOME Shell isn't such an issue after all. I suspect we will see more of that with time.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 17:47 UTC (Sat) by sbergman27 (guest, #10767) [Link] (11 responses)

Gnome 1->2 was disruptive in the same kind of way that Bluecurve was. A simplification that the die-hards didn't like. And Sawmill was just crazy with options. Granted, the Gnome devs of the day went a little far. But whereas Gnome 2 opted not to provide, KDE4 couldn't provide, back when we discussed this before. The Gnome2 desktop adapted to users more than the users adapted to the desktop.

OK. So I'll make another prediction. Gnome-shell will adapt or die. File that away and hold it as a hole card for some future year. As a pessimist, I'm generally pretty happy to be proven wrong. ;-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 18:06 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (10 responses)

That is a pretty lame prediction. It essentially restates the basic premise of evolution.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 18:25 UTC (Sat) by sbergman27 (guest, #10767) [Link] (9 responses)

Rahul, Rahul... I'd have expected better. You predict that users will adapt to Gnome3, and claim they have adapted to KDE4. I predict that Gnome3-shell will have to adapt to users, rather than the other way around, and make no claim regarding KDE4, at this time. That's not a restatement of the basic premise of evolution. I do agree that something akin to the principles of biological evolution are involved. I disagree with your particular application.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 18:32 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (8 responses)

You seem to have trouble following. Let me break it down for you. KDE users who were upset by KDE 4 seemed to have embraced it much more with incremental improvements in KDE 4.x. Now you can argue that it is KDE 4.x that adopted to users and not the other way around as well but the end result is just the same. More users using an desktop environment they weren't quite happy with before.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 19:18 UTC (Sat) by sbergman27 (guest, #10767) [Link] (7 responses)

Keeping in mind that I'm making no particular claims about KDE4's history after we last spoke about it, long ago... you now have a desktop environment which maintains the name "KDE", and a set of current users. I was a fan of a desktop named KDE back in the late '90s. It morphed a bit throughout the 2000's. And then disappeared. Are you saying that a desktop named KDE is back with more or less the same set of users? Does it matter if they are the same? Or will any old set of users do? I'm not being facetious here. Does it matter, do you think, if you please existing users, or dump them and find others? It matters to the users, of course. But should it matter to the developers?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 19:32 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (6 responses)

I think it should and atleast to me, it does. You can sometimes trade existing users for new users especially if you think you have a shot at capturing a new segment earlier on but often, your long term users engage with you a lot better than new users. Bug reports from users are much as a validation for me as kudos are because I love the fact that users are using the software I helped develop or maintain. This is especially true if you are voluntarily working on anything.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 20:08 UTC (Sat) by sbergman27 (guest, #10767) [Link] (1 responses)

Very nice post, Rahul. Very honest. Probably a good time for me to stop and say that I have always respected you and your work, and consider you to be an Internet friend, of sorts. I only spar as a hobby. ;-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 20:24 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link]

Thanks. I don't argue with people as much as I argue about technology and I enjoy doing that because it challenges notions and gives everyone an opportunity to learn and of course, it is nothing personal. I will be happy to ack that I agree with anything if I do.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 22:09 UTC (Sat) by Wol (subscriber, #4433) [Link] (3 responses)

:-)

That's why I happily transitioned from KDE3 to KDE4.

I gather a lot of KDE's troubles were because KDE 4.ZERO was pushed onto users, when the devs were quite open that ".0 status means the API is frozen", not that KDE4 was ready for real use.

Cheers,
Wol

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 16, 2013 21:38 UTC (Sun) by rahulsundaram (subscriber, #21946) [Link] (1 responses)

That is the usual rationale cited for the 4.0 release but it is clear that many users didn't expect that and 4.0 release announcement didn't mention it either. That was one of the reasons for the initial backlash and 4.1 announcement did include such a note. Chalk it up to lesson learned.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 13:47 UTC (Mon) by nix (subscriber, #2304) [Link]

In hindsight release 4.0 should have been the one with the codename "Krash", not 3.9something.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 16, 2013 23:51 UTC (Sun) by pboddie (guest, #50784) [Link]

To be fair, only the most enthusiastic had KDE 4.0 pushed onto them. Indeed, some of us have avoided KDE 4.x almost completely until now. Apparently, KDE 4.x is quite usable provided that one is prepared to adjust it somewhat to behave as one might expect.

As I pointed out elsewhere, perhaps the most significant problem for the KDE and GNOME developers is not what these environments can do or support but how they are delivered to users by default, especially when those users expect something else and are not willing to experience a learning curve for the sake of it (maybe because they're only getting version upgrades infrequently, not at every opportunity, and thus experience the resulting big paradigm change as a sudden shock).

Still, I think it is regrettable that only as various environments reach their x.7 release or so (where x is the controversial major version number) are they regarded as picking up from where the previous major version series left off.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 19:29 UTC (Sat) by hadrons123 (guest, #72126) [Link] (1 responses)

@rahul
Disclaimer:
If you think I 'm personally attacking you, I am really sorry if it came out that way. Its not my intention. But I always liked you for your contributions in fedora and massively respect your involvement. But the LWN "gnome-is-always right/everyone loves gnome-shell" posts of yours is misleading. I am using fedora f19 with XFCE.

>KDE 4 has settled down and users have accepted it.

If there was a choice people shall opt for it. Shoving down the throat with an not-so-interesting interface is what gnome users are facing right now.
You always argue that gnome UI is loved by everyone and try to project that the people who doesn't like it are a minority. Well honestly its not the case and you tend to use the freedom of speech to high-pitch your opinions on others on every other gnome-shell issue case. I always find some gnome-devs who doesn't disclose their position coming into support gnome-shell as well, at LWN. For all the love you have with gnome-shell try googling "I hate gnome-shell" without the quotes. There are like 2 dozens of threads in every linux forums about how they hate gnome-shell. I hardly find as much threads about how "I love gnome-shell" anywhere even with gnome 3.8.
There is very good chance that you might argue that forums are wrong place to look for statistics or info. I do understand that, but there are not many options out there.

>3.8 apparently has convinced some users that GNOME Shell isn't such an issue after all.

Do you have any base for your assertion?
If people don't talk about it, either they are done talking or already moved on to something else.
If gnome-shell is not such an issue why red hat is opting for a classic mode? (please save your self some time of implying how gnome-shell classic mode is also the gnome-shell, we already know that fact.)

> I suspect we will see more of that with time.
You are expecting us to get convinced and not bitch about gnome-shell?

>That is a pretty lame prediction. It essentially restates the basic premise of evolution.
I have lost lot of features in gnome apps in the last 2 years and I had to switch to XFCE. If that's evolution so be it.
There are alternatives in fedora as well with MATE/cinnamon.

>sbergman27 : Gnome-shell will adapt or die.
Its already adapting to red hat with classic mode. You can't market modern gnome-shell to workstation guys. Red hat knows that very well. They refer that "they want continuity in interface for a customer moving from gnome 2 to gnome 3".

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 19:37 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link]

"But the LWN "gnome-is-always right/everyone loves gnome-shell" posts of yours is misleading"

Let me stop you right there. I don't think GNOME Shell or GNOME is always right at all. They have made a lot of decisions which they themselves recognize as wrong and reverted and some I still think they have a long way to go but when I see people pretending that some change is universally hated or has no chance at all, I step it to point out, that isn't the case (be it KDE 4, GNOME 3 or Anaconda UI) and I am willing to take the heat for it. I don't expect that "bitching" about anything will solve any of your problems but hey, it is a free world.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 16, 2013 13:57 UTC (Sun) by paulj (subscriber, #341) [Link] (11 responses)

GNOME 1 to 2 was disruptive because GNOME 2 had very different code, and was initially quite buggy. The UI framework was unchanged though, other than that it focused on simplicity.

Further, the UI changes in GNOME 2 came about *BECAUSE OF* systematic, semi-scientific HCI testing, initiated by Sun, which led to a coherent HIG for GNOME. The GNOME people had objective *EVIDENCE* that the GNOME 2 UI changes significantly improved things.

I've asked here several times before, where are the HCI studies that justified the GNOME 3 UI changes? Not yet received a pointer to any such studies.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 16, 2013 17:49 UTC (Sun) by rahulsundaram (subscriber, #21946) [Link] (10 responses)

GNOME 1 to GNOME 2 issues wasn't about a differences in codebase or stability as much as it was about the substantial UI changes and all the different configuration options in GNOME 1 that didn't exist in GNOME 2 and most of which never came back. As for HCI studies, I suspect you already know the answer. It was a one off thing funded by Sun because they wanted to replaced CDE.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 13:16 UTC (Mon) by paulj (subscriber, #341) [Link] (9 responses)

That HCI studies on GNOME 1 were funded by Sun for some reason in no way answers the question:

Where were the HCI studies on GNOME 2 that provided the objective evidence and rationale for the GNOME 3 UI changes?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 16:52 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link] (8 responses)

HCI studies in GNOME 1.x won't explain GNOME 3 changes anymore than it would explain KDE 4 changes. GNOME 3 changes were driven via the current GNOME Design team.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 16:56 UTC (Mon) by paulj (subscriber, #341) [Link] (7 responses)

Are you being deliberately obtuse, or do have such a great compulsion to have the last word that you must reply even with such ridiculous off-the-point answers?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 17:18 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link] (6 responses)

I gave you a pointer to the team driving the changes so that you can ask them about the process directly. Did you find that too obtuse for you?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 20:10 UTC (Mon) by paulj (subscriber, #341) [Link] (5 responses)

It's a really simple question, with no reference to GNOME 1 or KDE required:

Where were the HCI¹ studies on GNOME 2 that provided the objective evidence and rationale for the GNOME 3 UI changes?

There's no need for a politician-like evasive answer, just "I don't know of any" or "Here's the link: ..." will do.

1. Or any other systematically obtained data or evidence.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 20:19 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link] (4 responses)

Let me try again for the last time. I am aware of some usability studies but I don't know if they have been published online. If you really want to know, you should talk to the people involved rather than asking here randomly and hoping the relevant designers will see and answer you.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 20:21 UTC (Mon) by paulj (subscriber, #341) [Link] (3 responses)

Ah, so now you know of studies, but can't tell me anything about them. Can you give me the contact details of a person involved in such a study so I can contact them?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 20:45 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link] (2 responses)

Again, you are insisting on asking the wrong person. You have to talk to the GNOME designers involved.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 21:04 UTC (Mon) by paulj (subscriber, #341) [Link] (1 responses)

Oh come now. Yesterday you didn't seem to know of any studies, except the Sun one on GNOME1 that led to the GNOME2 HIG:

Posted Jun 16, 2013 17:49 UTC (Sun) by rahulsundaram
… As for HCI studies, I suspect you already know the answer. It was a one off thing funded by Sun

Today you seem to claim you do know of some relevant to the GNOME2 → GNOME3 changes. So, as you have only just learned of them, you must have this information close to hand. Why be so unhelpful as to refuse to pass along a more exact pointer to something that surely must be almost at your fingertips?

In other comments in this article you seem willing to go into detail about and/or are quite confident you understand: what the design decisions were for the GNOME3 UI; what you have heard from the GNOME designers; why RedHat fund GNOME; etc. Why suddenly would you become so coy on the evidence question?

As of this point, there is still no answer to my question:

Where were the HCI studies on GNOME 2 that provided the objective evidence and rationale for the GNOME 3 UI changes?

with any pointer to any objective evidence.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 21:12 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link]

You are confusing the context of what I said and jumping to conclusions. You were talking about the Sun HCI study in GNOME and I pointed out that it was a one off thing funded by Sun because they replaced CDE. I honestly don't have any further information to provide to you.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 21, 2013 19:26 UTC (Fri) by strycat (guest, #91546) [Link] (2 responses)

KDE 4 is still a long way away from being the working DE that KDE 3 was. We've only "accepted" KDE4 because Gnome (2, 3, whatever version) still sucks more than KDE4 and the "Trinity" fork seems to still be just one or two people working on it.

Many of us recognize that because KDE is less popular than Gnome it just doesn't have the manpower to have good viable forks and alternatives grow. Gnome on the other hand has legions of coders who have made everything from Gnome Classic to Gnome Cinnamon.

So for us KDE people we're stuck with either using the inferior KDE 4, switching to the even more inferior Gnome, or go with something that is being maintained by just one person.

I accept these are the choices, but please don't say we've accepted KDE 4.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 21, 2013 20:39 UTC (Fri) by dlang (guest, #313) [Link] (1 responses)

There are a lot of us who have accepted KDE 4.

As someone who's been running it for several years before 4.0, I have to say that I don't know what is missing from KDE 4 that was there in KDE 3

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 25, 2013 19:27 UTC (Tue) by nix (subscriber, #2304) [Link]

Multi-key shortcuts? (Sure, not hugely important, but I have them in Emacs so I want them on my desktop! Also I have huge fanout of literally hundreds of shortcut keys and I don't think I can fit them into the keyboard with single keys without using up all my bucky bits just on KDE shortcuts, leaving none for Emacs.)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 18:36 UTC (Sat) by hadrons123 (guest, #72126) [Link] (3 responses)

Do you have any valid statistics to disprove it?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 18:55 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (2 responses)

Let me get this straight. You claim that everyone hates it without providing any reference to back it up and now you want to provide proof when I disagree? Alright. Let's do that.

To disprove your claim, all you need is a sample of one but you can go beyond that very easily. Look at Fedoraforum polls on which UI majority users preferred.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 19:23 UTC (Sat) by sbergman27 (guest, #10767) [Link] (1 responses)

Well, I dislike Gnome3. And I'd certainly never foist it on my users. I'm not sure Fedora Forum is the right place for such a poll. Fedora users, by definition, are content with alpha-grade software.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 19:28 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link]

That is hilarious. As someone who has dealt with bug reports across hundreds of packages, let me assure, Fedora users expect and demand a lot more than you assume.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 6:12 UTC (Mon) by russell (guest, #10458) [Link] (18 responses)

Users speak, nothing happens. Redhat speaks and all of a sudden we have classic mode. What more proof do you need. Nobody else could get there ideas into the GNOME club. They were just shouted down. Sorry but that's the way it is and a reason why I stopped using GNOME years ago. If you don't like that people leave "show them the code" but don't criticise them for not using your product.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 14:52 UTC (Mon) by raven667 (guest, #5198) [Link] (17 responses)

You don't think Classic mode is a result of feedback from users, do you think that the idea came down on stone tablets from on-high via Redhat? This whole conversation is ridiculous, GNOME creates a well-executed but divisive new UI, they slowly complete it and flesh it out while listening to feedback from users and implement a Classic UI mode to cater to those users, the Internet explodes in sarcastic flames about how they have failed to be true to their vision ... LOL WUT? Would everyone prefer they didn't take feedback and criticism and improve? If the Classic UI was some sort of Redhat-only fork (which they could certainly do) and not an official part of stock GNOME then maybe you'd have a leg to stand on but clearly enough GNOME developers are interested in a Classic UI that it has happened.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 15:50 UTC (Mon) by sbergman27 (guest, #10767) [Link] (2 responses)

"""
and implement a Classic UI mode to cater to those users
"""

It's really too bad we don't have user polls available here. The Gnome3 shell and Fedora folks have managed to alienate most of Gnome's user-base. Sure, one can point to current Fedora users as a source of people who are not quite so averse to Gnome3 shell. But Fedora's user share has, itself, dropped precipitously since its inception of Gnome3 shell.

I would never have guessed, 5 years ago, that the entire Linux desktop effort would self destruct in just a few years. Unity. Gnome3 shell. It's a wasteland out there today. Gnome 2.32 is as excellent as it ever was. And the Mint/Cinnamon/Mate guys are doing their best with limited resources.

On Unix/Linux servers, everything's a file. But aside from that, on the Linux desktop, everything's a phone.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 3:18 UTC (Wed) by jmorris42 (guest, #2203) [Link] (1 responses)

Yup. And that is where I suspect GNOME 3 came from. A few years ago the iPad dropped on the world, Microsoft was about to unleash Metro and the GNOMEs got a real bad case of tablet envy.

What a difference a few years make. Apple didn't turn the Mac into an iProduct, Microsoft is watching Windows 8 become the biggest failure since Windows ME (Bob is still their biggest bomb though.) and GNOME3 utterly failed to convince anyone it was a viable desktop. I was asking a long time ago whether RedHat would be dumb enough to try passing it off on their actual paying customers. Now we know the answer, no they aren't.

With a little luck, perhaps the madness that befell our industry is passing. Microsoft is restoring the Start button and if they don't provide an actual menu behind it other certainly will. Now we know it will be at least RHEL8 before GNOME Shell could be inflicted on corporate Linux desktop users and by it might not even exist. People might finally be realizing that desktops and tablets are not interchangable. Ubuntu still hasn't figured it out, but they are trying really hard to get on tablets so maybe Unity will work for em.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 20, 2013 17:02 UTC (Thu) by nim-nim (subscriber, #34454) [Link]

My suspicion is more 'Yuck, Nokia dropped GTK for QT! Intel is investing in clutter! The ingrates! We'll prove them wrong by making a better interface than them!' (plus a lot of wishful thinking like 'fixing evolution bugs is hard, let's make a browser launcher and let Google take care of the problem. The people that get interviewed do JS and CSS.')

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 22:19 UTC (Mon) by Zizzle (guest, #67739) [Link] (1 responses)

I think you are rewriting history here.

GNOME 3 was released in 2011 and we are now all of a sudden seeing a working Classic UI.

Not so long ago GNOME devs were talking about dropping fallback/classic entirely (yeah, because everyone loves shell so much).

You can see how people would think that RedHat is far more influential to GNOME than actual users right?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 22:31 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link]

Developers always have more influence than users on the direction of the project but "fallback/classic" is a confusing thing to say because they aren't the same thing. GNOME developers wanted to drop fallback mode because a number of components were starting to depend on clutter and wouldn't work properly in fallback mode. Fallback mode also required maintaining additional components like GNOME Panel. Classic mode doesn't have those issues.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 1:47 UTC (Wed) by russell (guest, #10458) [Link] (11 responses)

Yes it came down on stone tablets from Redhat. There was huge reaction from users, but nothing happened. They were just labelled malcontent. Take a look at https://fedoraproject.org/wiki/Statistics total connections to repositories. Fedora lost almost 1/2 of it's users when it switched to GNOME 3 and it has not recovered. Redhat does not want that to happen to RHEL.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 2:00 UTC (Wed) by rahulsundaram (subscriber, #21946) [Link] (9 responses)

"Fedora lost almost 1/2 of it's users when it switched to GNOME 3 and it has not recovered. Redhat does not want that to happen to RHEL."

You should be very careful with asserting number of users. The statistics page you are linking to makes this disclaimer

"Currently, there is no reliable way to determine the total number of Linux users, or even count the total number of users of any Linux distribution which does not have a mandatory per user registration process."

Fedora has also changed how it tracks connections so you will have to have a lot more qualifiers to add. Besides, a substantial number of RHEL customers don't run GNOME at all.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 3:56 UTC (Wed) by hadrons123 (guest, #72126) [Link] (3 responses)

> Besides, a substantial number of RHEL customers don't run GNOME at all.

Why do you bring RHEL here? Its totally irrelevant to Fedora statistics.

> Fedora has also changed how it tracks connections so you will have to have a lot more qualifiers to add.

Like a million? Maybe.
Still the usage statistics are low.

I would say significant number of Fedora/Ubuntu users ran away to Linux mint and Arch Linux.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 4:10 UTC (Wed) by rahulsundaram (subscriber, #21946) [Link] (2 responses)

" Why do you bring RHEL here? Its totally irrelevant to Fedora statistics."

It appears you missed the context. I am replying to a post that talks about both RHEL users and Fedora statistics.

"Still the usage statistics are low."

Fedora statistics on that page cannot track usage or users but only unique IP connections directly made to the public mirror manager for updates. Nothing more. There are dozens of different ways that these numbers can be undercounted or overcounted. It can vaguely show some general trends. Don't try to read too much into it.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 4:25 UTC (Wed) by hadrons123 (guest, #72126) [Link] (1 responses)

Well even if its under-counted it can't possibly be more than a million. If its over-counted its still low usage.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 4:37 UTC (Wed) by rahulsundaram (subscriber, #21946) [Link]

Maybe but I think you missed the point which is that these numbers are quite fuzzy and as I have noted in another post, mirror manager changes could very well explain the difference. Infrastructure knew that these were fuzzy numbers and Fedora Board asked Legal if a yum feature that enabled more accurate tracking could be enabled by default and the answer was that, it was risky due to international privacy laws and it was a costly process to quantify that risk and Fedora as a project decided we didn't want to do that. Even assuming that there is a significant change in these numbers, it is pretty hard to pinpoint the reason for that. Fedora as a distribution certainly doesn't try to judge usage by these numbers. Fedora project is planning to run a survey shortly which might help provide some answers.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 4:04 UTC (Wed) by russell (guest, #10458) [Link] (4 responses)

That disclaimer may hold for determining absolute numbers but look at the trend leading up to fedora 14 and after fedora 14. A 50% change is not noise. Forums are not noise. A lot of people won't or can't use GNOME 3.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 4:24 UTC (Wed) by rahulsundaram (subscriber, #21946) [Link] (3 responses)

Again, I am not sure that conclusion can be made from the statistics page. For instance, IIRC around that time, Fedora switched from a direct mirroring system to a tiered system and doesn't get as much load on the master mirror as a result and it might affect the IP count as well.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 15:43 UTC (Wed) by hadrons123 (guest, #72126) [Link] (2 responses)

If that metric is so bad, why still show it on the wiki?
Hosting it on the official wiki with a small disclaimer is poor marketing.
It should be taken down, immediately since it is giving such a bad rapport to the distro.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 16:51 UTC (Wed) by rahulsundaram (subscriber, #21946) [Link]

It has a rather big and prominent disclaimer. You just chose not to read it.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 17:14 UTC (Wed) by jspaleta (subscriber, #50639) [Link]

It's a poor marketing tool because its _not_ a marketing tool. Marketing had nothing to do with these numbers.

In an effort to inhabit your headspace, and to have some empathy for your point of view...I'm going to go ahead and horribly misuse an analogy similar to the way you are misusing these numbers in a sincere effort to communicate with you at your level.

Stop trying to use a turkey baster to put gas in your car. Turkey basters are not gas cans... in the same way these stats are not market penetration numbers.

The small unfeeling reptilian part of my brain would be overjoyed if Fedora was doing a mandatory phone home to get solid numbers for client usage..like Canonical introduced in 2010 in oem pre-installs when it started to require the canonical-census package be installed and active. Which interestingly enough we've never actually seen the data from that market penetration tool publicly discussed or even cited as part of marketing materials before it was taken out back and shot. Funny that. Or not so funny that depending on your sense of humor I guess.

But the evolved human part of my brain, the part that cares about other people and not just calculation accuracy, balks at the idea of tracking fedora clients. Just because we have the ability to track, does not make it ethical to do so. Oh yes something like a fedora-census application that was installed by default and pinged the fedora mothership every day would be an absolutely fantastic market analytics tool and at the same time be an absolutely horrid affront to user privacy. And with that trade-off in mind, I'll live with the ambiguous fedora unique ip numbers and the untrendable nature, thank you very much.

That being said, I've invested a non-zero amount of time trying to squeeze useful information out of the fedora numbers as well as other public datasets. Getting a handle on any real-world usage of any linux distribution is an impossible problem at the moment. Made worse by the fact that default user agent strings in firefox and chrome don't list linux vendor any longer. There is a reason why "unknown linux" is the highest linux population in the wikimedia stats for over a year now. The default user agent strings are just not unique enough any longer to see the difference between active linux desktop releases. Everything comes up as "unknown" because the vendor is no longer typically encoded in the useragent string. Wikimedia will count active opensuse and fedora release clients as "unknown linux" unless a user delibrately changes the default user string. I've know of no statistically significant way to trend a specific distribution flavor in day-to-day sampling interactions let-alone attempt to trend relative penetration of one distro to another. Every single set of numbers I've seen have gaping head wounds in the viability of the methodology to provide sensible estimates.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 19, 2013 8:45 UTC (Wed) by kigurai (guest, #85475) [Link]

The "Total connections to repository" is on the other hand the only statistic on that page that agrees with your assumption.

Also notice that the repository count took a just as big skydive between Fedora 8 and 9 as well.

Looking at the unique IP's to connect to fedoraproject.org there is nothing that indicates that Fedora has lost users. Just to take another metric as "counter-proof".

If you try hard enough, anything can be shown by statistics. :)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 8:54 UTC (Fri) by nim-nim (subscriber, #34454) [Link] (19 responses)

Red Hat is not a charity, when it contributes to a project (and it contributes heavily to GNOME) it expects to use it in some form in its paid offerings. I doubt GNOME could exist in its current form without Red Hat sponsorship. This announcement is probably the gentlest possible way it could remind the project it must sort its user story quickly. Red Hat will not dump its desktop customers so GNOME has better to include them in its target user base or it will have to go without a big sponsor.

As was expressed euphemistically the GNOME people heart is elsewhere.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 10:00 UTC (Fri) by ovitters (guest, #27950) [Link]

Classic mode is upstream.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 12:47 UTC (Fri) by AndreE (guest, #60148) [Link] (17 responses)

If Red Hat are "behind" GNOME as you have said above, to the extent that they have "veto" power, then why do they need to send a message to the GNOME team in the first place?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 20:17 UTC (Fri) by cyperpunks (subscriber, #39406) [Link] (16 responses)

Let's try: RH drops all GNOME (and related) development for 18 months. I doubt the project will survive.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 20:26 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link] (10 responses)

Let's try what exactly? No company is going to drop development just to answer this.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 18:17 UTC (Sat) by sbergman27 (guest, #10767) [Link] (9 responses)

Sure, you can work on that, that, and that if you want. But we're only going to pay you for work you do on this, this, and this. Billion dollar publicly owned corps drop projects all the time. And it's easier for Red Hat than for most. It's to their credit that that tend *not* to do that. But they certainly can.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 18:57 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (8 responses)

And they won't. So this experiment isn't going to happen. Sorry.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 19:37 UTC (Sat) by sbergman27 (guest, #10767) [Link] (7 responses)

How do you know they haven't? I've noted Red Hat doing a lot of things I'd not have expected. The number of ties at Red Hat Summit multiplies yearly. Like rabbits. The market-speak becomes more and more common. It's getting harder and harder to find sessions which aren't thinly veiled marketing ambushes. How can you be so sure? If they can coerce otherwise self-respecting employees to show up at a public event wearing ties, they can coerce them to do anything... for a paycheck.
Do they? Do they not? Will they decide to in future?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 19:40 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link]

I don't expect them to drop GNOME development just to satisfy the curiosity of cyperpunks on whether GNOME will survive without Red Hat. I don't think there is much to argue about that, honestly.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 22:26 UTC (Sat) by raven667 (guest, #5198) [Link] (5 responses)

What's wrong with wearing ties? I happen to like looking good in a smart shirt and tie from time to time. I've never understood the fascination with jeans, t-shirts and trying so hard to look schlubby. In my shop where most people wear shorts and sandals, showing up fully dressed is almost a subversive act. 8-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 13:53 UTC (Mon) by nix (subscriber, #2304) [Link] (3 responses)

> What's wrong with wearing ties?

Some studies suggest that they increase the risk of strokes and aneurysms by up to 40%. They definitely increase cranial blood pressure, which seems like an extremely unwise thing to voluntarily do to yourself to me.

More to the point, they're uncomfortable and don't even look very nice. Why *are* people still wearing modified 16th-century Hungarian (or was it Czech?) war garb anyway?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 14:28 UTC (Mon) by raven667 (guest, #5198) [Link]

> Some studies suggest that they increase the risk of strokes and aneurysms by up to 40%. They definitely increase cranial blood pressure, which seems like an extremely unwise thing to voluntarily do to yourself to me.

Can that be expressed in cases per 1000 or similar? That kind of statistic is totally uninformative, misleading even, without knowing the magnitude of the values involved. It might be that chances of death by tie are still substantially smaller than getting hit by a bus, for example.

> More to the point, they're uncomfortable and don't even look very nice. Why *are* people still wearing modified 16th-century Hungarian (or was it Czech?) war garb anyway?

I believe it is Croatian and that the term cravat is a corruption of that. The fact that the practice has survived is potentially evidence that people like it. I think it looks good. Why fit in when you were meant to stand out 8-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 18:58 UTC (Mon) by dlang (guest, #313) [Link]

> More to the point, they're uncomfortable and don't even look very nice.

some people are uncomfortable wearing ties, other people are not.

looking nice or not is a matter of opinion.

For many people who wear ties (and a large number who wear suits regularly), they get ones that are comfortable to wear and feel uncomfortable in a business setting when they are in other attire.

don't mistake your opinion for facts.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jul 1, 2013 5:56 UTC (Mon) by elvis_ (guest, #63935) [Link]

That's easy to fix, get a shirt that fits around the neck. If your tie is throttling you then something is wrong.

(my contribution to the longest thread in history, humble though it may be)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 14:57 UTC (Mon) by sbergman27 (guest, #10767) [Link]

I think it's a fair bet that Red Hat management looked better in tanktops, shorts, and sandles in 2000 than they will in 2014. (Only a select few of us get better with age.) I'm not advocating for that dress code. Only that they not dress as if for a funeral.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 16, 2013 1:42 UTC (Sun) by AndreE (guest, #60148) [Link] (4 responses)

Right, the other companies and projects that rely on GNOME related technologies would simply sit on their hands. Not to mention the individuals within Red Hat who work on GNOME in their own time.

I guess Qt should be dying now after Nokia ceased backing it?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 16, 2013 23:59 UTC (Sun) by pboddie (guest, #50784) [Link] (3 responses)

Please don't bring Qt into this: for some people, not having the "backing" of Nokia means eventually being able to have official support for platforms that Nokia either didn't value or didn't like. Although some areas of the Qt ecosystem did benefit from Nokia's interest in the technology, one can also identify ways that the ecosystem was also damaged by having a narrowly focused custodian who eventually discarded the technology completely.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 9:03 UTC (Mon) by AndreE (guest, #60148) [Link] (2 responses)

Right, so you are proving my point. Losing an "official" backer obviously changes certain things, some for better, some for worse, but the health of the project is really determined by the size of the interested maintainers.

I think it is pretty ridiculous to believe that a project like GNOME and the associated projects that fall under the auspices of the GNOME Foundation would just wither and die if Red Hat as an organization bizarrely decided to stop supporting it's development. There are plenty of interested parties and I imagine a large collective of independent developers (many of which work at Red Hat in any case) who would continue to contribute, maintain, and drive the project.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 20, 2013 12:09 UTC (Thu) by pboddie (guest, #50784) [Link] (1 responses)

My point was that the "backing" of Qt by Nokia was a mixed blessing, contrary to some assertions. I agree that this makes it a bit like the GNOME technologies in the sense that work done externally and independently of Nokia - I'm thinking of things like the Android port, for example - kept it interesting and viable for certain groups of potential users, and in the case of GNOME, what Red Hat does with the technology can and does co-exist with what other people do.

But whereas Nokia's withdrawal from Qt development arguably led to renewed commitment from other parties (including significant commercial interests) and various other positive developments (the open governance initiative was finally completed), any Red Hat withdrawal from GNOME participation would prompt some serious introspection in the community because it would be yet another example of a large company walking away (Sun being probably the most significant case that comes to mind, but Canonical are also worth bearing in mind).

I suppose the spread of contributors is wider for the GNOME technologies and so that should make the community more robust than that around Qt, and there is therefore more hope that the community would weather that storm, but Red Hat doesn't have the arguably obstructive role that Nokia had which, upon being eliminated, opened up significant opportunities for those who might have been denied those opportunities with a single large vendor running the show.

In short, Nokia's exit probably had an identifiable positive side whereas such benefits are harder to envisage if Red Hat got out of GNOME development.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 20, 2013 13:38 UTC (Thu) by AndreE (guest, #60148) [Link]

fair point

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 12:33 UTC (Sat) by sbergman27 (guest, #10767) [Link] (5 responses)

RHEL, Centos, and SL... Islands of sanity in a landscape of total Chaos. Of course, Red Hat couldn't have allowed the Linux desktop to self-destruct. I feel ashamed for ever worrying that they might. The thing I need to confirm next is that 3D is not required for classic mode. I have a whole multi-city Linux business desktop infrastructure resting upon the excellent freenx-server platform. A 3D requirement is not an option for us. But with RHEL/Centos/SL 6, I'm good for another 7+ years, regardless.

Gods, I love Red Hat.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 13:24 UTC (Sat) by clump (subscriber, #27801) [Link] (2 responses)

Gods, I love Red Hat.

Wow, you've come a long way.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 13:51 UTC (Sat) by sbergman27 (guest, #10767) [Link]

What do you mean by that?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 16, 2013 16:11 UTC (Sun) by sbergman27 (guest, #10767) [Link]

I'm still curious what you mean by that. I've been unwaveringly impressed by Red hat as a company since Red hat 4.2. By that, I don't mean RHEL 4.2, but Red Hat 4.2, back in spring 1997. (Right before the tumultuous transition to glibc in Red Hat 5.0.)

I'm curious whether you are referring to some time at which I chose to (hopefully constructively) criticize some specific decision by Red Hat management. Or perhaps are conflating my views of Red Hat with those I hold regarding Fedora (which differ substantially). Or perhaps simply have me confused with someone else. (I noted another S. Bergman, posting as guest "sberg" here, the other day.)

-Steve Bergman

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 15:31 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (1 responses)

Classic requires 3D, which will be provided by llvmpipe if you have no hardware 3D.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 16:07 UTC (Sat) by sbergman27 (guest, #10767) [Link]

Oh shit. That sounds a bit scary from a WAN-transparency point of view. I need remote X to both work and remain performant over a WAN. Maybe this will do that and maybe it won't. At least RHEL 6.4 is supported until the end of 2020. I prefer potential flag days which are far, far away. ;-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 13, 2013 21:34 UTC (Thu) by atai (subscriber, #10977) [Link] (7 responses)

Yes, I cannot think how can a server OS accept a mobile style UI...

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 2:46 UTC (Fri) by imgx64 (guest, #78590) [Link] (6 responses)

To be fair, servers are usually installed without any GUI. I'm guessing they're talking about workstations.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 11:44 UTC (Fri) by dgm (subscriber, #49227) [Link] (5 responses)

Here we have a number of servers and all have GUIs. It makes Windows people' life less miserable.

I'm guessing you're talking about computing or storage nodes.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 12:50 UTC (Sat) by sbergman27 (guest, #10767) [Link] (4 responses)

RHEL is emphasizing gui tools, these days. e.g. if you want to manage LVM2 without a bunch of fiddly calculations on a calculator, and avoid worries about exactly what "-L +/- #G/g" really means in actual usable blocks or sectors, you need a gui. i.e. if you don't want to make a data-destroying silly mistake, you're best to have a gui on RHEL.

I've never understood why someone couldn't just write decent ZFS-style cli utilities for Ext4/LVM2, though. It would have been a hell of a lot easier than developing btrfs, and have provided a lot of the same benefits.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 13:32 UTC (Sat) by clump (subscriber, #27801) [Link] (2 responses)

I've never understood why someone couldn't just write decent ZFS-style cli utilities for Ext4/LVM2, though. It would have been a hell of a lot easier than developing btrfs, and have provided a lot of the same benefits.

LVM commands are pretty easy. There's not much difficulty in the following command example:

lvextend -r -L +10G /dev/vg_myvg/lv_root (bonus points if you follow the delightful -r option)

Red Hat has written great LVM docs, check them out here:

LVM Administrator's Guide

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 13:58 UTC (Sat) by sbergman27 (guest, #10767) [Link] (1 responses)

LV administration is pretty good. (And yes, -r is invaluable.) It's VG, PV, and partition management which I find lacking these days. Of course, I don't find complete coverage of those in the gui, either. It looks like the new dm-thin capabilities may be game-changing. And I've only cursory experience with that, at this time.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 13:57 UTC (Mon) by nix (subscriber, #2304) [Link]

> And yes, -r is invaluable.

I avoided using it of late because it relies on fsadm, which I thought had been deprecated. But now I look for it I see no sign of that anywhere. Maybe I was confusing it with parted, which has indeed discarded its fs resizing abilities.

I tend to just resize by hand using resize2fs. Often I don't even need to unmount to do an lvextend-and-resize2fs, which I think lvextend -r does require :)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 16, 2013 22:47 UTC (Sun) by dag- (guest, #30207) [Link]

I'd recommend looking into fsadm for all your LVM extending/reducing pleasure ;-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 13, 2013 21:58 UTC (Thu) by jhoblitt (subscriber, #77733) [Link] (15 responses)

I'm a little surprised to hear rhel7 plans to make xfs the default filesystem as it doesn't support shrinking. That's a nice feature to have for laptops/workstations although it's mostly irrelavant for large scale storage or VMs (typically easier to reprovision). The discussion of gnome 3 classic mode makes it clear they're not planning to abandon the workstation market... does this mean nobody cares about shrinking any more or perhaps that xfs support for that feature is closer than it appears from the TODO list?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 0:16 UTC (Fri) by dlang (guest, #313) [Link] (5 responses)

I think that it means that support for shrinking is less important than the advantages that XFS has for very large partitions.

It's not that shrinking doesn't matter, but that for an enterprise distro, large multi-TB or PB filesystem support matters more

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 0:23 UTC (Fri) by jhoblitt (subscriber, #77733) [Link] (4 responses)

There are numerous good things about xfs, however, if your root, /boot, etc. need to scale to several 100 TiBs your probably doing it wrong...

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 0:27 UTC (Fri) by dlang (guest, #313) [Link] (3 responses)

But if you are picking ONE default filesystem to use, it needs to scale

then you can get into the whole discussion on if it's really worthwhile to have lots of separate partitions or not, it's a nice bikeshed to paint ;-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 13:30 UTC (Sat) by sbergman27 (guest, #10767) [Link] (2 responses)

Not really a bikeshed topic anymore. At least I don't think so. I was die-hard "KISS" and advocated one big filesystem for years. But with enormous drives, LVM2, and now thin-provisioning, even I see the advantages of breaking things up a bit. Are you advocating a single fs? If so, we should probably agree, upfront, not to bikeshed too long. ;-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 22:14 UTC (Sat) by dlang (guest, #313) [Link]

my bikeshed comes in two colors :-)

on VM systems, I do a single filesystem for everything 'local' and then mount anything that will be shared

If I need more space, I create a new VM image, bring up the new image, failover to it, and destroy the old one

for bare metal servers, I create three filesystems
current root
future root
var

the OS goes in the current root, which I like to keep as small as possible (a couple of Gigs on most servers), everything else goes in var (with possible directory symlinks from root into /var). When I do a major OS upgrade, everything in /var gets moved to /var/var.old and then the new OS gets installed in the future root partition (which becomes current root). The post install configuration script can then pull any config info from the old installation, and if I need to revert, I just mv /var/* /var/var.new; mv /var/var.new/var.old/* /var change the bootloader to point back at the old root partition and everything is as it was before.

I am not a fan of LVM, it adds complexity to managing the system, if you actually use it you end up scattering your filesystem around the drive in ways that make it harder to optimize I/O, and most of the time I don't think it's benefits are worth it.

If you are dealing with a local drive where nobody else could use the drive, why leave any space unused?

SSDs are a possible exception, there is a reason to leave space unused, and scattering the filesystem doesn't hurt as there is no seek penalty. I'm just never had a SSD enough larger than I needed to consider it :-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 14:06 UTC (Mon) by nix (subscriber, #2304) [Link]

I became very glad that I'd been a multi-filesystemer when I had the big ext4 corruption extravaganza last year. Nothing like roasting your /home and /var repeatedly (a couple of times so badly I had to restore from backup) to make you very happy that most of your stuff, in particular most of that needed to boot, is on different, unaffected filesystems!

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 1:56 UTC (Fri) by ricwheeler (subscriber, #4980) [Link] (8 responses)

With thin provisioning in device mapper, you can think of shrinking as an obsolete feature.

Specifically, you can use a dm-thin target to give each user a large "virtually provisioned" file system (say 1TB? 4TB? whatever) and device mapper will add and remove storage from that virtual file system just like we do with virtual memory and DRAM.

I would also note that resizing a file system - even when supported like in ext4 - is almost always a horrible idea. It tends to skew the allocation of the data on disk in unnatural ways and performance can really suffer.

Last note, there are in fact XFS patches to support shrinking that have been debated on the lists.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 5:12 UTC (Sat) by neilbrown (subscriber, #359) [Link] (1 responses)

Would not dm-thin also tend "to skew the allocation of the data on disk in unnatural ways"?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 14:47 UTC (Sat) by ricwheeler (subscriber, #4980) [Link]

With device mapper thin targets, there is certainly the same danger of skewing the allocation under the file system. You can largely avoid this if you use a large data block size. The trick is to use a block size that works well for your snapshot needs and avoids as much as possible the potential for a performance hit.

We are working now to do some in depth performance testing at various block sizes with various workloads.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 11:59 UTC (Sat) by sbergman27 (guest, #10767) [Link] (5 responses)

So... how can I expect dm-thin to perform compared to a regular LV? I create the thin pool and dm-thin targeted LV, and format it ext4 in RHEL6.4. How well does that perform? I then create an hourly snapshot, so that if necessary, I can pull anything I want from an hour ago. How does the original dm-thin target perform? How does the snapshot perform? I then decide to create a snapshot for every hour of the last 12 business hours. How does the original then perform? How do the snapshots perform?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 12:02 UTC (Sat) by jhoblitt (subscriber, #77733) [Link] (2 responses)

What is the interaction of write barriers with dm-thin?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 14:52 UTC (Sat) by ricwheeler (subscriber, #4980) [Link] (1 responses)

Write barriers will work under dm-thin as well - it needs to propagate those barriers down the stack to the device under it in the stack.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 15:01 UTC (Sat) by sbergman27 (guest, #10767) [Link]

Will work? Do you mean "does work in RHEL 6.4"?

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 14:59 UTC (Sat) by ricwheeler (subscriber, #4980) [Link] (1 responses)

Performance depends on a lot of things - the storage you have at the base of your stack, the size of the block used in dm-thin, your file system and of course, your specific IO pattern.

I would encourage you to test with your combination and share the results with us on the upstream lists :)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 15:21 UTC (Sat) by sbergman27 (guest, #10767) [Link]

And here I was, tired of my recent benchmarking, and trying to take the lazy way out by just asking you on this Saturday morning. ;-)

On a Power Edge T310 with a 4 SATA drive md raid10 (yeah, the same one) if I benchmark my expected workload, how much confidence can I have in current RHEL 6.4 defaults? Would you expect the results to vary much between ext4 and xfs? Would you suggest any different starting points?

BTW, congrats on the move toward XFS as default RHEL fs. I was not expecting that. I'm still not certain whether I like that or not. But there is definitely something exciting about it.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 13, 2013 22:13 UTC (Thu) by wagerrard (guest, #87558) [Link] (52 responses)

I like Gnome Shell. When CentOS 7 is out, there is a very good chance I'll install it and switch it away from classic mode.

To each his or her own (RHEL users typically don't have that choice), but calling Gnome 2 the "traditional desktop" seems as useful as calling a 1966 Ford the "traditional car".

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 2:59 UTC (Fri) by louie (guest, #3285) [Link] (50 responses)

To be fair, there are a lot of enterprise customers out there who wish they were still running Ford '66, I mean Windows '98, because forward progress is expensive to them. So it is perfectly reasonable for Red Hat to cater to that market. But it is sad to see so many people in the development community applaud those dinosaurs.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 4:46 UTC (Fri) by billev2k (subscriber, #32054) [Link] (49 responses)

I may well be a dinosaur myself, but... I *remember* things differently than the new shells (Gnome 3, Unity) support. For an infrequently used tool or application, I remember "sort of where it was" and "sort of what it looked like". I can find it again in the application menu. I DO NOT remember what it was called, and can not find it again with "search".

It's the same with people. I remember faces, or what people do, or where I knew them from, far better than I remember their names.

So, for me these new shells are a disaster. Total stupidity. Useful information discarded for no reason. Try as hard as I may, I can't like them. Though I keep trying...

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 5:17 UTC (Fri) by bojan (subscriber, #14302) [Link] (44 responses)

But you don't understand... See, the shell has a notification system that hides notifications from you and desktop overview that is invisible, both of which will appear if you can _remember_ that there is something to remember. Kinda like that rememberall that Neville got in HP - he could never remember what it was that he had forgotten. ;-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 15:30 UTC (Fri) by drag (guest, #31333) [Link] (21 responses)

90% of what people bitch about can be solved by a extension. Which is what extensions are for because everybody likes to bitch about their particular pet peeve, and everybody's pet peeve is different.

ie:

If Gnome made notifications pop up then people bitch about notifications being distracting and worthless since they prefer to concentrate on one task at a time. (this is very literally a discussion that happened on LWN a while ago, and for a while quite frequently)

If you hide notifications so people can view them on their own time then people bitch that they could miss something important while they are busy concentrating on something less important.

With extensions you can have your cake and eat it too. In this way people that care about either method can have it without forcing cluttered configuration mess on the rest of the world that doesn't give a flip either way.

Extensions are not always a good solution

Posted Jun 14, 2013 21:14 UTC (Fri) by man_ls (guest, #15091) [Link] (8 responses)

For a long time Gnome advocates were trying to find sane defaults, and afterwards removing most settings altogether since nobody should need anything beyond what the system brings. Now the pendulum swings: the base system does not provide for common needs, they are provided by extensions which are:

hard to discover since they are not even included in the system,
completely non-standard so that desktops become incompatible for different workflows,
and may cause instability issues (thoughts of Mac OS 9 and its carnival of extensions make me shiver).

Both points of view have their merits: sane defaults and extensibility. I believe that there must exist some kind of sane compromise without going to extremes: a base system which is usable for everyone, and customizations which enrich the user experience.

Browsers are a good example. Mozilla devs don't say "we are not providing that setting since you can use extensions"; extensions serve niches such as developers, people with disabilities, and so on. 90% of user needs are not provided by browser extensions. If a sizable part of the population have some need and use a particular extension, then it will probably be integrated into the main codebase with a setting to enable it. I find it a good compromise.

Extensions are not always a good solution

Posted Jun 14, 2013 21:30 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link] (7 responses)

"Mozilla devs don't say "we are not providing that setting since you can use extensions"

Kind of amusing because extensions were created by Mozilla specifically because they removed a whole of bunch of features from the classic Mozilla browser when they created Firefox and wanted to find a way to deal with the backlash. Later on, it became an avenue for very popular and widely used add-ons like say AdBlock but also to add back some of the features that Firefox removed in recent versions including status bar and several others.

"If a sizable part of the population have some need and use a particular extension, then it will probably be integrated into the main codebase with a setting to enable it"

Unlikely this is the case since even in 2009, atleast 1/3 of all Firefox users were running with add-ons according to Mozilla and I bet they are never going to integrate features like AdBlock no matter how popular it is.

https://blog.mozilla.org/addons/2009/08/11/how-many-firef...

Another thing to note is that GNOME Shell has done some of exactly what you suggest here. To provide a specific example: the most popular extension in the GNOME Shell extensions site in the one to hide the ally icon by default and 3.8 makes it obsolete by not showing the icon and instead relying on just hotkeys to activate various a11y features.

Extensions are not always a good solution

Posted Jun 14, 2013 21:46 UTC (Fri) by man_ls (guest, #15091) [Link] (5 responses)

Unlikely this is the case since even in 2009, atleast 1/3 of all Firefox users were running with add-ons according to Mozilla

Not really a contradiction: even if 1/3 of users have some extension, the set of extensions is very diverse and there is probably no extension with even 10% of users. Not even adblock which is probably the most popular -- and as you hint will never be integrated with the browser for political reasons.

Extensions are not always a good solution

Posted Jun 14, 2013 22:00 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link] (4 responses)

Firefox extensions used by users do have a long tail but there are several obviously very popular extensions that don't get integrated into Firefox for various reasons. Adblock isn't the only one. It is just the most widely known example. Take a look at

https://addons.mozilla.org/en-US/firefox/extensions/?sort...

Video DownloadHelper isn't going to be integrated into Firefox either nor is Greasemonkey despite their popularity.

Extensions are not always a good solution

Posted Jun 14, 2013 22:20 UTC (Fri) by man_ls (guest, #15091) [Link] (3 responses)

Interesting. Adblock Plus boasts 15,523,890 users. Even with its dwindling 27% market share in Europe (pop. 739,165,030), and a penetration rate of the intertubes at just 50%, that would yield about 100 million Firefox users. Those are low estimates: the penetration rate at least would be well above 50% in most countries since many people have two or more computers.

So if only European users existed the market share of Adblock would be 15%. Taking into account the whole world in our silly little exercise its market share must be well below 10%, unless the Firefox figure is grossly misrepresented. I have checked with my own locale and the figure stays the same so I believe it should be the global total. The FAQ says that this info is collected by Firefox daily unless you opt-out, so the figure should be reliable.

tl;dr: even the most populars Firefox add-ons are well below 10% of the global user base.

Extensions are not always a good solution

Posted Jun 15, 2013 3:55 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (2 responses)

Those numbers are useful as a general indicator but I wouldn't really rely on those numbers to draw up %'s. Distributions routinely disable the automatic check. It works only when you download binaries directly from them.

Extensions are not always a good solution

Posted Jun 15, 2013 8:12 UTC (Sat) by man_ls (guest, #15091) [Link] (1 responses)

The point is anyway that Mozilla or Chrome do not usually reject popular functionalities on usability grounds, or because they do not fit with some "vision". I find it very healthy.

Extensions are not always a good solution

Posted Jun 15, 2013 14:26 UTC (Sat) by khim (subscriber, #9252) [Link]

Both Chrome and Mozilla rejected many things on because they do not fit with some "vision". Think ActiveX (which was implemented in Firefox, but was never enabled) or RSS tags (which are not supported by Chromium), etc. And that's just the tip of the iceberg.

Extensions are not always a good solution

Posted Jun 17, 2013 15:53 UTC (Mon) by nye (guest, #51576) [Link]

>Kind of amusing because extensions were created by Mozilla specifically because they removed a whole of bunch of features from the classic Mozilla browser when they created Firefox and wanted to find a way to deal with the backlash.

That's not correct. Mozilla was extensible long before Phoenix came about - people were using extensions to add things like tabs and gestures, for example.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 21:20 UTC (Fri) by khim (subscriber, #9252) [Link] (2 responses)

90% of what people bitch about can be solved by a extension. Which is what extensions are for because everybody likes to bitch about their particular pet peeve, and everybody's pet peeve is different.

Right. Failure mode number 3a: make it so configurable the fact it has no sane defaults is now somehow your fault.

It's still a failure mode, sorry. Some configurbility is good, but not does not mean defaults should be unusable.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 22:49 UTC (Fri) by pboddie (guest, #50784) [Link] (1 responses)

Indeed. It isn't a virtue that each time someone adopts Linux on the desktop they have to have expert help to make everything sane. I guess that's where Red Hat comes in, however.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 16:03 UTC (Mon) by sbergman27 (guest, #10767) [Link]

"I guess that's where Red Hat comes in, however."

The entire Gnome/Fedora/Red Hat relationship brings to mind the old TV cliche that "Parental supervision is advised". ;-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 3:34 UTC (Sat) by bojan (subscriber, #14302) [Link] (8 responses)

> 90% of what people bitch about can be solved by a extension.

FF: extensions enhance the product. Gnome 3: extensions fix the product. :-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 3:49 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (7 responses)

There are some that can arguably be called fixes in either product and brings back features dropped in a new release by default. The extension to bring back status bar in Firefox and the top icons extension in GNOME Shell are examples of this but the large majority of them in both Firefox and GNOME Shell are really enhancements that target a specific audience and cannot ever be shipped in the core product at all and I use quite a lot of them. Noscript in Firefox and Transmission indicator extension in GNOME are examples of the latter.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 15:42 UTC (Sat) by sbergman27 (guest, #10767) [Link] (3 responses)

Ever the apologist, Rahul. Perhaps gnome3-shell was just ill-conceived? That's a valid possibility, you know. And Red Hat did effectively nix it. I doubt it will be around for RHEL8. (The name might be. But the form won't.)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 16:42 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (2 responses)

Red Hat nixed the GNOME Shell it funded because it shipped with a bunch of extensions? Sure. That makes a lot of sense and hey, good luck with another one of the random predictions.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 17:24 UTC (Sat) by sbergman27 (guest, #10767) [Link] (1 responses)

They certainly nixed the UI "vision" of the Gnome3 shell guys. I don't think anyone doubts the solidity of the underlying libraries. But you make a good point. We'll have to see if the resulting desktop is usable, or if it turns out a zombified mockery of RHEL6's desktop.

Which random prediction are you referring to, btw? Sometimes I'm right on those. But usually I'm wrong. I take it you remember one where I was wrong. Let's here it, and all have a laugh. :-)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 18:11 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link]

Calling one among two options in the default display manager as "nixed" shows a flair amount of taste for dramatization. However, there is an often overlooked point hidden in there. The feedback from commercial customers is important.

As to your predictions, I will just point out that if you have a lot of influence or control over a software component or distribution without the strong feedback loop from customers, you end up with Ubuntu and Unity as opposed to RHEL and GNOME.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 15:48 UTC (Mon) by nye (guest, #51576) [Link] (2 responses)

>The extension to bring back status bar in Firefox

I've read a couple of people mentioning this, but I'm confused; perhaps somebody can clarify?

So far as I can see from trying FF21 with no extensions, there *is* a status bar, but it autohides if there is nothing to show. That is, if it would otherwise be empty, then it's not shown. So are these extensions there purely to make the status bar show all the time, even when it's empty?

If that's the case, what reasons are people giving for needing to see the empty status bar? This seems weird which is why I wonder if I've misunderstood.

(Having spent an inordinate amount of time in around 2005 trying to get the status bar to behave as it does now may be colouring my interpretation of the situation.)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 16:56 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link] (1 responses)

Maybe this https://addons.mozilla.org/en-US/firefox/addon/status-4-e... can answer you.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 10:14 UTC (Tue) by nye (guest, #51576) [Link]

Oh, the progress bar!

All of the text that would once have been in the progress bar does still show up in the current status bar ('transferring from blah', etc), but there's no actual progress indicator, which appears to be the principal complaint.

That does help, thanks.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 16:32 UTC (Fri) by kigurai (guest, #85475) [Link] (21 responses)

How is the activities overlay invisible?
There is basically nothing else to press other than the "Activities" button when you start GNOME.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 3:38 UTC (Sat) by bojan (subscriber, #14302) [Link] (20 responses)

You are right. I forget about the telepathic interface. My bad. Sorry. :-(

I just tried it now. I stared at Activities button for 1.45 seconds. It showed me where all my windows were on all workspaces. It's like magic!

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 4:02 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (12 responses)

Perhaps there is some cross talk here. When you talk about notifications and overview being hidden, it appears to the reader as if they are in the same category but they aren't. Activities is an obvious place to access the overview. Calling it hidden doesn't make much sense.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 4:16 UTC (Sat) by bojan (subscriber, #14302) [Link] (11 responses)

So, you switch to workspace 5 and start a long running task. Then you forget about it and go back to whatever you were doing before. What tells you that there is a long running task? In Gnome 2 (and other similar environments), it is your workspace switcher. In Gnome 3, nothing (unless you can _remember_ that you need to go looking for something). What's the point of the _graphical_ user interface if not to give you _visual_ clues as to what is going on?

Same with notifications. They appear, they disappear. Absolutely no clue there is anything to take care of unless you can _remember_ there is something to look for.

I wonder how many times people will log out and kill the long running task or not handle hidden notifications because of this basic oversight. It is the computer that needs to work for me, not the other way around.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 4:34 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (10 responses)

"So, you switch to workspace 5 and start a long running task. Then you forget about it and go back to whatever you were doing before. What tells you that there is a long running task?"

I never really run into this problem at all since I have always used alt+tab to switch between windows and GNOME Shell shows the windows from all workspaces. I had to use the workspace switched in GNOME 2 because alt+tab in GNOME 2 didn't behave that way.

On the other hand, when I deployed GNOME 2 on a large scale, users who accidentally moved the windows in a different workspace never really found it and the very tiny indicator in the workspace switcher didn't help at all. They would launch multiple instances and complain that Linux was slow.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 4:45 UTC (Sat) by bojan (subscriber, #14302) [Link] (9 responses)

So, you mass deployed multiple workspaces to people that were not comfortable (or familiar) with this and when they got confused it was their fault? OK.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 4:56 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (8 responses)

Where did I blame anyone? Just pointing out that the default setup of GNOME 2 didn't really solve the problem you are talking about (ie) good indication of applications running in other workspaces. You have to remember to look at the workspace switcher and that was hardly obvious.

GNOME 3 atleast attempts to solve the problem in several ways:

a) the overview shows all the workspaces (which are dynamic, so your eyes are drawn to it if you have more than one) and apps running inside it in a much more visible way compared to the switcher.

b) alt+tab shows applications from all workspaces

c) clicking on a running task in the dash just changes focus to the already running instance instead of launching another one so you can always do the same thing and not worry about whether an existing instance is there or not.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 9:35 UTC (Sat) by bojan (subscriber, #14302) [Link] (7 responses)

> Just pointing out that the default setup of GNOME 2 didn't really solve the problem you are talking about (ie) good indication of applications running in other workspaces. You have to remember to look at the workspace switcher and that was hardly obvious.

So, you don't educate your users on what facilities are available, you don't customise their systems to either disable this behaviour (i.e. configure a single workspace) or enlarge the switcher (higher panel) so that it is more visible and then claim it's not good because "it's not obvious". Well, it sure is more obvious than a _hidden_ workspace switcher.

And then there is the "default setup" claim. Hey, in Gnome 2 you could actually customise things using configuration knobs - which is almost completely impossible in Gnome 3. The new system doesn't even have a central configuration for its main elements on the panel. It's all just cowboy code from whichever extension you happen to have. Yeah, sooo much better.

It is not me saying that people's short term memory is bad. It is usability scientists that claim that:

http://www.nngroup.com/articles/short-term-memory-and-web...

Why on earth would anyone want to hide the whole of the desktop view from users is beyond me. The shell design documents treat users as neurotic idiots that have nothing better to do than click on buttons just because they are there. So, what do do? Remove the lot to overview.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 17:00 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (6 responses)

UI research has shown the majority of users tend to stick with the defaults. By default, the workspace UI in GNOME 3 is much more obvious than the GNOME 2 switcher because the overview shows all these elements together. Now one could spend a lot of time training the users and fiddling with the knobs but at that point, GNOME Shell is much more customizable and you lose the argument entirely.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 6:19 UTC (Mon) by bojan (subscriber, #14302) [Link] (5 responses)

> UI research has shown the majority of users tend to stick with the defaults.

In your particular case the bad ones you have configured for them.

> Now one could spend a lot of time training the users and fiddling with the knobs but at that point, GNOME Shell is much more customizable and you lose the argument entirely.

You are being hilarious - the sysadmins are just dying to write JS. Gnome shell cannot even do a simple drag and drop on the panel - something version 2 could do many years ago. As I said before, it's all just cowboy code of whichever extension combo you happen to have loaded, instead of the actual ability to customise.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 7:32 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link] (4 responses)

I didnIt was the default gnome 2 configuration which was not very usable . Sysadmins would install extensions as needed. There is no need to write any code

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 0:19 UTC (Tue) by bojan (subscriber, #14302) [Link] (3 responses)

Your original statement:

> when I deployed GNOME 2 on a large scale

Then later you claim:

> I didnIt was the default gnome 2 configuration which was not very usable .

So, if you were the one that deployed Gnome 2 on a large scale and failed to provide the appropriate default configuration (which is more than possible in Gnome 2), this is then the shortcoming of Gnome 2. Great.

You know, if you said that you deployed a few systems and defaults were shit, I'd say fine. But you didn't. You deployed large scale (you were obviously _the_ sysadmin in charge), so it was your responsibility to deploy sane defaults. In all the places I worked thus, sysadmins do that as a matter of course.

> Sysadmins would install extensions as needed. There is no need to write any code

Oh really? So, if I decide (as a sysadmin) that I'd like to have the panel on the bottom (because, I don't know, everyone is used to that in the legacy system) and the extension for that does not exist (I didn't actually check - but any number of such examples can be constructed), how exactly will that be done without writing JS? It cannot be done in Gnome 3, because Gnome 3 does not have a configuration database for the elements of its panel (unlike Gnome 2, which can also have multiple panels, for instance).

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 0:42 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link] (2 responses)

GNOME 2 default workspace experience doesn't solve the probem you are talking about and yes that is obviously the fault of GNOME 2.

" Oh really? So, if I decide (as a sysadmin) that I'd like to have the panel on the bottom (because, I don't know, everyone is used to that in the legacy system) and the extension for that does not exist - but any number of such examples can be constructed), how exactly will that be done without writing JS?"

In this case, an extension obviously does exist and is part of GNOME Classic however assuming there are enhancements that you want that isn't part of the desktop environment, writing new code isn't the responsibility of the system administrators but just in case they have to, extending GNOME Shell is much more easier than GNOME 2.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 1:09 UTC (Tue) by bojan (subscriber, #14302) [Link] (1 responses)

> GNOME 2 default workspace experience doesn't solve the probem you are talking about and yes that is obviously the fault of GNOME 2.

Still trying to shift your responsibility to others.

> In this case, an extension obviously does exist and is part of GNOME Classic

You are thinking of the wrong panel.

> writing new code isn't the responsibility of the system administrators but just in case they have to, extending GNOME Shell is much more easier than GNOME 2.

I really love it when you pretend that you don't understand, when I know that you do, given that you are a Fedora contributor.

You will remember the accessibility icon problem in the early Gnome 3. Code had to be written to remove it. Any such action by a system administrator now (i.e. Gnome 3.8) that does not have an extension written for it will require code writing (which then may conflict with a number of other extensions - who knows). No such problems exist in Gnome 2, because elements are laid out using central configuration.

Ease of code writing is irrelevant for this exercise. Lack of central configuration database for panels is.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 1:16 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link]

The default GNOME 2 experience is the sole responsibility of upstream developers and not anyone else. Extensions do have a configuration database, ordering and preferences now and they can all be controlled via Gsettings and gnome-shell-extension-prefs

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 9:07 UTC (Mon) by kigurai (guest, #85475) [Link] (6 responses)

Uhm, if you press "Activities", which is the only button that you can really press, then all your windows and workspaces are visible on the overlay.

If you prefer the old workspace switcher in the panel, then that is fine (believe there is an extension for it), but don't try to sound like the workspaces and windows are in some way hidden.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 9:17 UTC (Mon) by johill (subscriber, #25196) [Link] (1 responses)

Nope, it's still completely broken with multiple screens :-(

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 13:33 UTC (Mon) by kigurai (guest, #85475) [Link]

I have had no problems using dual screens. I know that there are differing opinions on how the extra screen should be used and behave though.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 0:28 UTC (Tue) by bojan (subscriber, #14302) [Link] (3 responses)

This is from the Nielsen's Windows 8 usability study:

> "When users can't view several windows simultaneously, they must keep information from one window in short-term memory while they activate another window. This is problematic for two reasons. First, human short-term memory is notoriously weak, and second, the very task of having to manipulate a window—instead of simply glancing at one that's already open—further taxes the user's cognitive resources."

Now extrapolate that to workspaces, lack of workspace switcher and the need to click on Activities button and get overview in order to _see_ (this is what I call zero visibility problem of the shell).

The fact that someone wrote an intellihide style extension that is the workspace switcher just proves that the original ideas from the shell design documentation are very dubious. The point of a _graphical_ user interface is to enable to you to _see_ things - not to hide them.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 7:41 UTC (Tue) by kigurai (guest, #85475) [Link] (2 responses)

You could just as well extrapolate that quote to imply that workspaces is a really bad idea since you can't always see all windows.

And that is why extrapolating is seldom very useful.

I think we understand that you personally don't like G3. And I think we are all fine with that. What I fail to understand is why you spend so much time trying to convince us who think that G3 is really nice and works well, that we are somehow wrong and should go back to G2.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 10:57 UTC (Tue) by Cyberax (✭ supporter ✭, #52523) [Link] (1 responses)

That is actually why MS and Apple are moving away from virtual desktops.

However, virtual desktops have a redeeming feature - they are a boon for "power users", because they typically assign each virtual desktop a fixed role and can efficiently switch between them.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 12:00 UTC (Tue) by kigurai (guest, #85475) [Link]

I agree, and for me personally this works great with Gnome 3.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 6:19 UTC (Fri) by xtifr (guest, #143) [Link] (1 responses)

Mouse to or click on Activities, click Applications, and the menu is on the right. It's displayed totally differently from traditional menus, but it's still there, and it still basically works the same.

There's even a shell plugin to make it appear next to Activities as a regular menu on the top bar (Advanced Settings->Shell Extensions->Applications Menu Extensions).

It may be because I was expecting the worst after hearing all the horror stories about how bad it was, but I've actually been quite pleasantly surprised after trying Gnome Shell. It's definitely odd, but it really doesn't seem to be that much worse (or better) than what it replaced.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 10:15 UTC (Fri) by darrylb123 (subscriber, #85709) [Link]

I actually found that Gnome 3 worked better for my workflow. I have a small number of applications that I use all day, all the time. Each is in my shortcut menu. I installed the extension that allows you to set which application goes to which desktop, eg my mail client is always in the top desktop and my terminals are always in the third ...
Now I just Ctrl-Alt-up or down the relevant number of desktops to get where I want to be. When I get lost, then I use the mouse to the top left corner and I am away again.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 13:21 UTC (Sat) by sbergman27 (guest, #10767) [Link] (1 responses)

The problem with both Gnome3-Shell and Unity are that they are solutions in search of a problem. If it ain't broke don't fix it. Gnome2, and the desktop metaphor have lasted so long not because of inertia, but because they've served us damned well. New ideas are not necessarily better ideas. In this case it is not the users who are perseverating on Gnome2. It's the Gnome3-Shell devs who are perseverating on their own failed ideas.

Red Hat had no choice but to slice through the crap and act based upon reality. "Modern", indeed. Now there's an amusing euphemism. (*chuckle*)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 20:41 UTC (Sat) by luya (subscriber, #50741) [Link]

"It's the Gnome3-Shell devs who are perseverating on their own failed ideas."
Extensions in Gnome Shell is a failed idea?
The ability to mimick the look similar to Gnome 2 is a failed idea?
Setting to "Have File Manager handling Desktop" through Tweak Tool or dconf-editor is a failed idea?
Please explain.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 13:25 UTC (Fri) by robclark (subscriber, #74945) [Link]

yeah, I like gnome-shell/modern too.. even more so when I try to go back to a gnome2 desktop and try to use it. I'm more productive w/ gnome-shell, therefore it is better for me. Just throwing that out there, because it seems like the only vocal ones in this debate are the ones afraid of change ;-)

gnome-shell classic mode seems like a good compromise to bring the best of gnome-shell to rhel users without surprising them too much out of the box. The modern mode is there for users to try if/when they want. Classic mode isn't a sign of failure of gnome-shell/modern.. but IMHO it shows the flexibility of gnome-shell design, and it gives a less abrupt transition for users.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 13, 2013 22:23 UTC (Thu) by dowdle (subscriber, #659) [Link]

Videos from Red Hat Summit:

RHEL Roadmap Part 1 - http://videos.cdn.redhat.com/2013-summit-platform-2.mp4
RHEL Roadmap Part 2 - http://videos.cdn.redhat.com/2013-summit-platform-3.mp4

Ogg format files available as well.

See all of the videos by selecting the various sections:

https://www.redhat.com/summit/2013/gallery/

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 13, 2013 22:26 UTC (Thu) by dowdle (subscriber, #659) [Link] (18 responses)

Given the fact that the GNOME 3 Shell is really designed for hardware accelerated videocard setups... and most servers don't have those... who in their right might would make it the default for a distribution primarily targeted at servers?

Of course if you have the appropriate videocard setup, then you should be able to do fancier desktops.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 13, 2013 23:05 UTC (Thu) by geofft (subscriber, #59789) [Link] (8 responses)

Do people actually have _any_ desktop environment on servers? I can't recall the last time I had one.

I assume RHEL is considering enterprise desktops, not servers.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 13, 2013 23:20 UTC (Thu) by dowdle (subscriber, #659) [Link]

There are some graphical tools that are handy to run on servers sometimes... so while I usually don't have X11 running all the time on a server, I frequently have it installed.

Your point is taken about enterprise desktops... but you have to wonder what percentage of RHN subscriptions are for those.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 8:29 UTC (Fri) by nim-nim (subscriber, #34454) [Link]

RHEL is not only servers, it's also workstations (granted, there are not many companies that have lots of unix workstations anymore, but there are some).

And even if you don't have any use for workstations, you still need some unix desktop nodes to control your unix servers (because some idiot isvs have decided the cli was pre-windows past, so their server sofwtare has a java gui frontend. The easiest way to handle those is to install a minimum X stack on the server and do an ssh -X from a workstation running a similar linux gui stack). The most painful is to control your servers from windows + exceed (some companies did try this to kill all unix desktop remnants when desktop unix=solaris; it didn't go well…). If wayland works well as rdp server it will probably see huge adoption in enterprises just to kill the X workstations needed today to run unix GUI tools remotely. RHEL 8 may not have much need for GNOME in any form.

IIRC one another point raised during this roadmap was that RH customers were sick of the server/desktop segmentation and wanted a single distro that did both. (another point that totally contradicts the "desktop and server are different" "common sense" that justified many GNOME3 decisions).

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 12:58 UTC (Fri) by ajmacleod (guest, #1729) [Link] (5 responses)

I maintain a graphical desktop environment on a (CentOS) server - it's a "terminal server" so would be rather useless without it!

It provides remote desktops via traditional XDMCP and NX and works very nicely indeed; Gnome 2 is reasonably easily managed system-wide (through GConf mandatory and default keys). This aspect could certainly be improved, but this "traditional" desktop is what virtually every company actually wants.

They don't want fancy animations, they don't want fancy compositing features; they want pretty much exactly what Gnome 2 provides.

My own E17 desktop is completely different and would doubtless be a nightmare for other people to have to use; but most people in most non-IT businesses want something that looks familiar (Traditional if you wish) and provides an uncomplicated way of launching and switching between web browser/mail client/word processor.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 16:48 UTC (Sat) by sbergman27 (guest, #10767) [Link] (4 responses)

"It provides remote desktops via traditional XDMCP and NX and works very nicely indeed;"

Yes. I maintain ~100 remote user desktops in 4 cities using NX. Quite amazing, isn't it? There is no way we're moving to Gnome3 shell. I'm just hoping that classic mode doesn't destroy performance. I'd like to have the option to upgrade to RHEL7. But if we can't we can't. A Mate option would be nice. Even if it comes from rpmforge. I might even accept it from EPEL if it came to it.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 17:48 UTC (Mon) by ajmacleod (guest, #1729) [Link] (1 responses)

It's amazing to me that this type of setup is not more commonly used - as you say, it works really well. I do worry that even the possibility of this kind of configuration has been sort of forgotten about by many developers though...

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 22, 2013 21:02 UTC (Sat) by jmorris42 (guest, #2203) [Link]

Well one reason to avoid it is longterm viability. It is pretty clear that remote X isn't considered a useful feature by the people working to replace X. And virtual machines aren't important either since none support GL yet all 'modern' desktops now require it. And no, LLVM isn't an answer if you want to do more that confirm installation is possible with it.

And they want to replace X because... well mostly to replace X because replacing things that work with new shiny things that don't is the way to the Year of Linux on the Desktop someday in the far future. But all this churn has actually done is ensured none of the years in the near past or future ever had a chance at widescale adoption. Because one or more major systems is always in a state of disrepair.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 17:53 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link] (1 responses)

EPEL includes Xfce FYI

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 18, 2013 19:22 UTC (Tue) by ajmacleod (guest, #1729) [Link]

XFCE is a nice efficient desktop and I used it a few years back in a previous server fulfilling the same role; however I found Gnome2 settings much easier to manage system-wide through mandatory / default gconf keys.

A quick search shows that XFCE does appear to have gained something similar in xfconf so perhaps I will consider it again for a future build (hopefully not any time soon!)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 13, 2013 23:28 UTC (Thu) by DOT (subscriber, #58786) [Link] (4 responses)

Only the animations are a problem when you have no hardware acceleration; compositing itself is not a very expensive operation. So, get rid of the Activities animations and you have almost 0 CPU usage. But since the animations are a tiny part of a mostly idle session, you might as well leave them on.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 2:30 UTC (Fri) by bojan (subscriber, #14302) [Link] (3 responses)

Have you actually watched the CPU usage with Gnome Shell over llvmpipe (animations disabled)? Do yourself a favour and try it. It is anything _but_ zero.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 12:07 UTC (Fri) by SEMW (guest, #52697) [Link] (2 responses)

Tangentially related: I've been running gnome-shell for a little while (on my desktop), I never thought to check its CPU usage. I have a reasonably fast desktop, I guess I assumed it'd be negligible.

Nope. 10%. Constantly. As a baseline. Plus another 5% for X, which is presumably also due to gnome-shell. (And yes, glxinfo shows OpenGL hardware acceleration as working correctly, using the nvidia binary driver and not llvmpipe).

That's... insane.

That's for 3.6. I'll try upgrading to 3.8. If that still uses 15% CPU as a baseline, I think I'm off to XFCE.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 12:48 UTC (Fri) by AlexHudson (guest, #41828) [Link] (1 responses)

Are you sure that's not just a problem with your set-up?

I have a decent laptop w/Intel drivers, and gnome-shell rarely appears to use anything in the process list. It gets to maybe 1% now and then, if I cycle between the desktop and overview continually it chews a bit of CPU but that's kind of abnormal.

Compared to the other stuff I have running, it's just noise.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 17:32 UTC (Fri) by SEMW (guest, #52697) [Link]

I've just updated to raring and installed gnome-shell 3.8 from the gnome3-team ppa, and... the problem's gone. gnome-shell now idles at ~0-2%, which is fine & expected. So chalk it down to a now-fixed bug, I guess.

gnome-shell 3.8's also quite a bit nicer than 3.6 in other ways (it's actually responsive - I don't have to wait a second after pressing winkey before the animation starts - and the new TopIcons extension lets you have a useful taskbar), so I'm going to stick with gnome-shell.

Cheers :)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 5:39 UTC (Fri) by Frej (guest, #4165) [Link] (2 responses)

Classic is based on Shell. It is not gnome 2.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 15:40 UTC (Fri) by drag (guest, #31333) [Link] (1 responses)

I think that Redhat took the 'fallback mode' and made it non-shit. Probably wrong though.

It will be interesting to see how well it works.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 17:55 UTC (Sat) by pbonzini (subscriber, #60935) [Link]

Wrong. :) No offense intended, of course.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 16:01 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link]

GNOME classic mode is just a bunch of extensions. It requires hardware acceleration as much as the regular mode does.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 11:52 UTC (Fri) by dgm (subscriber, #49227) [Link]

Now that's a sensible choice. Make new stuff available, but don't make it the default.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 14:35 UTC (Fri) by joshl (guest, #91369) [Link]

I would have been surprised if Red Hat decided to do anything else. If anything, it's a testament to the usefulness of the extension system the Gnome folks have implemented.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 17:18 UTC (Fri) by amacater (subscriber, #790) [Link] (9 responses)

Upgrades between versions is the BIG new thing for Red Hat - and old hat for anyone that's run Debian for the last 15 years ... :)

Desktop environment - pointless on a server, as stated, until you want to run Java / GUI tools / you run something that needs an Apache project that demands a webserver or five (tomcat/jetty/apache ...all on the same machine).

Much more important - if ext4 is now well supported, what's the maximum partition size - it needs to be more than the current 16TB in Red Hat Enterprise Linux 6.4 (backward compatibility with ext3? It can be annoying).

Anaconda/installer changes _will_ break people's default assumptions - I can't remember if this is also the release which switches X.org to VT1 rather than VT7 ...

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 14, 2013 22:23 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link] (8 responses)

" Upgrades between versions is the BIG new thing for Red Hat - and old hat for anyone that's run Debian for the last 15 years ... :)"

That isn't really true. Red Hat has provided an option for upgrades between versions for as long as RHEL existed via Anaconda "upgradeany" option. The article is talking about a fully commercially supported upgrade path that presumably includes maintaining compatibility for third party ISV apps. Debian has never done that and cannot do that since they don't offer any commercial support in the first place and I doubt cares about ABI compatibility across versions either.

" Much more important - if ext4 is now well supported, what's the maximum partition size - it needs to be more than the current 16TB in Red Hat Enterprise Linux 6.4 (backward compatibility with ext3? It can be annoying)."

https://www.youtube.com/watch?v=AbXFJMiCCkE (25:00) suggests that XFS will become default for RHEL 7 and upto 50TB for Ext4 will be supported by Red Hat.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 19:09 UTC (Sat) by amacater (subscriber, #790) [Link] (7 responses)

OK - I'll bite. The recommended upgrade procedure between any Red Hat Enterprise Linux major version amd the next is to reformat and reinstall: nothing else is supported as a matter of course. I don't have many servers I've done this on but there is at least one that has gone from Debian 4.0 - 7.0 without a full reformat on the same hardware.

ISVs don't care enough about the packages they produce for Linux, in my experience. That's one of the reasons why there are so many -compat packages, old versions of OpenSSL required and poor dependency management. The likelihood of an ISV coping from RHEL 4.6 to 6.4 is small unless they bundle everything statically in their application. Oh, and Debian really does care about ABI breakage - but sometimes the world breaks round you.

I recognise the name and reputation behind it, however, so am happy to defer to your greater knowledge and expertise.

XFS as default - may not be ideal for something I'm considering, which is more than a little peeving. I'm still not sure to do for >60TB as a filesystem with a mix of file sizes and access requirements. XFS still doesn't quite feel as well known as Ext3/Ext4 and Ted Ts'o :)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 19:26 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (2 responses)

The reason why upgrades haven't been recommended before by Red Hat despite the fact that upgradeany technically works fine (I have done it or even an yum upgrade for that matter) is because of ABI compatibility concerns and commercial support burden. While voluntary supported distributions certain can care about packages within its repository, I don't think it deals with any issues for third party applications or kernel modules and I don't fault them for it. I wouldn't if I wasn't getting paid to do it either. It certainly gives you a lot more flexibility when you control everything and can rebuild from source whenever necessary.

ISV's and especially their users deeply care about their software not breaking due to upgrades and they will do whatever ugly hacks it takes (compat libraries, bundling, static linking, building off an ancient version of RHEL. using only %post in RPM to shove files into place after using their custom build environment .. you name it)

It is true that Ext* is more recognized within Linux community but XFS has seen quite a lot of deployments especially in the high end market. I don't doubt that it will work out well. Unfortunately (but not very unexpectedly) it has taken Btrfs a long time to mature but Red Hat in the summit presentation has said they intend to offer commercial support for that as an option (along with Ext* using the Ext4 codebase and XFS) as well.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 19:59 UTC (Sat) by sbergman27 (guest, #10767) [Link]

"""
It is true that Ext* is more recognized within Linux community but XFS has seen quite a lot of deployments especially in the high end market. I don't doubt that it will work out well.
"""

At the very least, XFS will see some much needed testing in areas where it has traditionally not seen much use. Red Hat has to think way ahead. High end today is home user in 10 years. Or less. Though I'm finding it really hard to utitilize the space and I/O bandwidth provided by modern drives. It's the damned random writes that I have ongoing issues with.

Random prediction. I might be eating my own words in a few years' time. But if RHEL were not the first to make XFS default, who would be? Fedora, I suppose. This is all very interesting.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 17, 2013 2:30 UTC (Mon) by jhhaller (guest, #56103) [Link]

I was quite surprised by the Red Hat kernel manager saying that they weren't sure about the use case for btrfs, and thought it would be desktop. Our current fileserver environment consists of ZFS on OpenSolaris because of the snapshot capabilities and what they enable for backups and redundant systems. I would think this would be a good use case for btrfs, but not if Red Hat plans to target support to desktop.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 15, 2013 22:02 UTC (Sat) by dlang (guest, #313) [Link] (3 responses)

> XFS as default - may not be ideal for something I'm considering, which is more than a little peeving. I'm still not sure to do for >60TB as a filesystem with a mix of file sizes and access requirements. XFS still doesn't quite feel as well known as Ext3/Ext4 and Ted Ts'o :)

I think it's worth pointing out that Ext3/4 do not get tested on 60TB disk arrays because the developers (including Ted) just do not have such systems available to them.

I don't remember the exact bug, but there was an Ext4 bug a year or so ago that only happened on large filesystems, and when he was explaining why they didn't catch it in testing, Ted explained that they do the vast majority of their testing on laptop/desktop type systems, almost all with just a single drive for the filesystem.

It's been the case for a long time that if you needed top performance on a system with lots of drives, XFS would give you better performance than ext*

As "Big Data" becomes more common, even with larger drive sizes, use of large arrays increases.

In addition, XFS has been improving it's performance on smaller drives, and addressing some of it's historic bottlenecks (creating or deleting large numbers of files for example)

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 16, 2013 2:19 UTC (Sun) by sbergman27 (guest, #10767) [Link]

"""
The developers (including Ted) just do not have such systems available to them.
"""

One day, Google may grow large enough to be able to supply such hardware for testing.

One interesting thing that is happening is that storage *space* is increasing by leaps and bounds. But performance (particularly seek performance) isn't. Hasn't for a very long time. The number of terabytes per user is increasing. But a home user is still a home user. And sure, they can have a multi-terabyte DVD image collection. But they only watch them 1 by 1, at 30MB/s or whatever.

For stuff that matters, even on most server hardware, ext4 is still the best. 16TB is still a freaking huge amount of space. And we should not forget that. Ric asked at Red Hat Summit 2013, how many of their enterprise user attendees were above 16TB, and seemed surprised at the relative lack of hands shown.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 22, 2013 21:04 UTC (Sat) by kleptog (subscriber, #1183) [Link] (1 responses)

For one project we were actually testing XFS, until we read about xfs_repair requiring 1GB of memory for every TB of disk. With a 96TB array on a machine with 64GB of memory we got nervous. So we ran xfs_repair and yup, it OOMed. Adding swap just made it take forever, not really making progress. So we're back to ext4.

Now, googling manpages for xfs_repair now shows an option for limiting memory usage so maybe this is fixed now, but there are still plenty of pages on the web warning about this.

Red Hat discloses RHEL roadmap (TechTarget)

Posted Jun 22, 2013 22:31 UTC (Sat) by dlang (guest, #313) [Link]

> but there are still plenty of pages on the web warning about this.

just because a problem gets fixed doesn't mean that all the pages that reported the problem get changed as well. some of the people who wrote the pages will write new ones, some won't bother, some won't know the problem got fixed.