Who is that code for?

Posted Jul 1, 2011 22:48 UTC (Fri) by clugstj (subscriber, #4020)
In reply to: Who is that code for? by hadess
Parent article: Who is that code for?

It's not that the focus of GNOME is not what (some) people want that is the major problem. The big issue, for me at least, is the supreme arrogance that you get in response from the lords of GNOME. Their attitude seems to be that users are so stupid that they cannot be allowed to have options - even the ones they had before.

Who is that code for?

Posted Jul 1, 2011 23:28 UTC (Fri) by mjg59 (subscriber, #23239) [Link] (24 responses)

Options cost. There's the extra code behind them. There's the extra complexity in the UI that the user needs to interpret. There's the combinatorial explosion of unintended behaviours resulting in bug reports that are difficult to track down. If you can reduce the number of options without making it impossible for people to work, you're making life better for everyone.

The problem is that if you ask people whether they need an option, the answer is likely to be "yes" even if said person is actually entirely capable of managing without it. People don't like to voluntarily change. You don't have a good idea of what people really need until you cut down on the available options and force them to. And sometimes in the process of trying to do that you end up trimming off more features than is ideal, and you end up with a situation like Gnome 2.0, where things perhaps did go to far and some functionality got re-added in 2.2 and 2.4. But after that pain, you have something that's less complicated and more usable than what came before it.

Gnome 1 -> Gnome 2 was a significant transition. 2 -> 3 is bigger in some ways and smaller in others. But many of the changes are similar in philosophy and the public response has strong parallels. The reason that some of the people involved seem arrogant is that they've been here before, done this before and it worked out fine last time round.

Who is that code for?

Posted Jul 1, 2011 23:37 UTC (Fri) by nix (subscriber, #2304) [Link] (6 responses)

People don't like to voluntarily change.

There's not much voluntary about this. People like involuntary change even less, which is why people seem to be fleeing GNOME in even greater numbers than they did in the GNOME 1->2 transition.

Who is that code for?

Posted Jul 2, 2011 0:22 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link]

"People like involuntary change even less, which is why people seem to be fleeing GNOME in even greater numbers than they did in the GNOME 1->2 transition"

I don't think you can ever know this for a fact. I can't begin to think you would be able to judge this even at a broad level from some comments from pretty much the same people in a few websites. Fedora 15 happens to be the first distro to include it as the default and forum poll showed opinions are divided but majority in fact liked it. Downloads have increased quite a bit compared to previous releases. That isn't definitive but we can only get some idea what the user base thinks in a year or so from now when the dust has settled down a bit.

Who is that code for?

Posted Jul 2, 2011 0:47 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (4 responses)

I don't know of any distributions shipping Gnome 3 that don't also have a non-Gnome 3 version that's still supported. Right now the change is voluntary. In the future you'll have to put more effort in if you want a Gnome 2-based setup, but 3.2 is also likely to have some of the rougher edges filed off and somewhat less offensive to the more traditional users.

Which isn't to say there won't be awkwardnesses. But I think that if you go back to the original 2.0 era, you'll find that people were pretty much as upset as they are now.

Who is that code for?

Posted Jul 2, 2011 3:38 UTC (Sat) by bronson (subscriber, #4806) [Link] (3 responses)

Fedora 15 dropped Gnome 2, right? Am I misreading this thread?

http://www.fedoraforum.org/forum/showthread.php?t=263445

Yes, I'm really glad Ubuntu still has it. For now...

Who is that code for?

Posted Jul 2, 2011 3:39 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (2 responses)

It did, but Fedora 14 is still a supported release.

Who is that code for?

Posted Jul 3, 2011 13:28 UTC (Sun) by Tet (guest, #5433) [Link] (1 responses)

Fedora 14 is still a supported release

Ha ha ha. Very funny. In theory that's true. In the real world, try logging a bug against anything but Rawhide or the latest released version, and it'll sit there until however many months are needed for it to be automatically closed for being too old. I like Red Hat. And Fedora is still my preferred distribution of choice. But it's getting harder and harder to defend that position with each new release, and I find myself looking around for alternatives. Sadly, outside of Ubuntu (which I wouldn't touch with a 10' barge pole), there's not a lot of choice.

Who is that code for?

Posted Jul 3, 2011 17:19 UTC (Sun) by rahulsundaram (subscriber, #21946) [Link]

Look at the amount of updates for Fedora 14 and it becomes very hard to see your statement as true. It is true that more things will get fixed in development branch or the latest version and that is true for any distribution

Who is that code for?

Posted Jul 2, 2011 0:30 UTC (Sat) by bronson (subscriber, #4806) [Link] (3 responses)

Worked out fine the last time? All's well that ends well, let's do it again?

Not sure I agree... I really resent the amount of time the Gnome team has forced me to waste on this 3.0 transition. First, to discover it doesn't work on my modern ATI laptop and file bugs. Second, to discover that neither my wife nor I get along the new UI (she says it's a computer trying to be a cell phone). And third, to not have a working fallback so now I need to learn LXDE or XFCE.

Is it too much to ask that a desktop environment be a little more conservative so I don't have to waste so much of my time on it? I have work to do. The Gnome 2 transition was never this bad.

Yes, I also believe this will work out OK in the end. But that doesn't mean that I ever want to go through it again.

Who is that code for?

Posted Jul 7, 2011 20:28 UTC (Thu) by ovitters (guest, #27950) [Link] (1 responses)

GNOME didn't force you anything.

Not want to be impolite, but it seems better to enjoy life than to keep posting the same argument on LWN.

Who is that code for?

Posted Jul 8, 2011 0:02 UTC (Fri) by bronson (subscriber, #4806) [Link]

I'm sorry for caring.

This argument is new. It's "yes, the transition really was that painful," not "focus follows mouse is being buried." Both points worth making, no?

Surprised you didn't apply your last sentence to yourself!

Who is that code for?

Posted Jul 8, 2011 3:02 UTC (Fri) by obi (guest, #5784) [Link]

That's interesting. I remember the Gnome2 transition to be far worse.

And maybe you'd like a desktop environment that's a bit more conservative - but I very much like the changes and continuous improvement. Like most people, 10% of the stuff rubs me the wrong way occasionally, but I file bugs, adapt, work around it, or wait it out on those things. And the other 90% improves my interaction with computers.

So yeah, I wouldn't mind them being even a bit more aggressive. But I do remember that I'm not the only (Gnome) user in the world.

Who is that code for?

Posted Jul 2, 2011 15:52 UTC (Sat) by paulj (subscriber, #341) [Link] (12 responses)

Options cost. There's the extra code behind them.

As a maintainer of some other software, this I can sympathise with. Options complicate the code, complicate testing, complicate life for the users, users will enable options that aren't appropriate for them, etc. Options absolutely need to be carefully controlled. "Just Do The Right Thing" is the much better option, whenever it is possible. Cause, as you say, otherwise this gets exponentially harder to cope with:

There's the combinatorial explosion of unintended behaviours resulting in bug reports that are difficult to track down.

However, if a fixed set of options make life hard in this way, then encouraging unfettered modification through code plugins is going be an order of magnitude worse. At least with options fixed in the code, the software developer trying to figure out a bug report will know the code concerned is "theirs" and always be able to find it.

I just can't square a project that on the one hand apparently wants to eliminate options because of how they complicate debugging user problems, while on the other hand encouraging users to address the lack of options by providing an unlimited plugin interface.

Who is that code for?

Posted Jul 2, 2011 16:27 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (11 responses)

It hasn't proven to be too much of a problem for the kernel, where bugs triggered while using out of tree modules are generally rejected until they can be reproduced without them. I'd hope that the gnome bug reporting infrastructure takes that into account.

Who is that code for?

Posted Jul 2, 2011 19:54 UTC (Sat) by bronson (subscriber, #4806) [Link] (9 responses)

Those situations are quite different, no?

The kernel discourages external modules and makes an effort to merge the ones even a few people use.

Gnome has been encouraging external modules and doesn't appear to have any intention of ever merging them. (going by what I read in the discussions)

Who is that code for?

Posted Jul 2, 2011 20:26 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (8 responses)

It's equivalent from the point of view of bug reporting.

Who is that code for?

Posted Jul 2, 2011 20:39 UTC (Sat) by nix (subscriber, #2304) [Link] (7 responses)

Only if you consider it acceptable to blow off all bug reports from users who happen to be using any modules. With a GNOME panel as featureless as it now is, that's not going to leave very many people who can submit bug reports that'll be accepted.

(This also would mean that if distributions install any modules by default, that they would have to commit to fixing all bugs in the panel that their users report themselves, without involving upstream. Since this is not going to be practical for the vast majority of distributions, it either means that the module system will be almost unused, or that distributions will have to throw away a lot of bug reports, or that GNOME *will* have to accept reports from module users.)

Who is that code for?

Posted Jul 2, 2011 20:46 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (6 responses)

Ensuring that a bug is reproducible without extensions being involved is a pretty basic part of tracking it down, and I think that's true of all projects that involve third party addons.

Who is that code for?

Posted Jul 3, 2011 20:28 UTC (Sun) by elanthis (guest, #6227) [Link] (5 responses)

His point is that he believes GNOME 3 is going to have a very high concentration of add-on users, where as most projects have a very low concentration of add-on users.

For instance, my kernel has at most ever had one non-standard module installed, for GPU support when the FOSS drivers outright failed to support the hardware. The vast majority of the time, my Linux kernels are pure. Bug reports them from me are not going to be rejected on grounds of taintedness. I'm willing to bet that most Linux users are in the same boat. If a sizable portion of GNOME 3 users are installing a half-dozen modules to fix an assortment of minor annoyances then a sizable portion of GNOME 3 users will not be able to submit bugs.

Especially when it comes to end-user software, it's valuable to move the effort onto the developers, not the end users. That in part is why Bugzilla is an abomination unto FOSS (go through fifty steps, including account creation and verification, to file "The about dialog spelled 'GNOEM' wrong"). Requiring users to create test cases because the developers are too lazy to do their damn job is also a problem. It's one that makes sense in an Open Source as a hobby world, but in a world where projects have foundations and paid marketing teams and are obviously backed by large commercial ventures, it's just wrong.

Who is that code for?

Posted Jul 3, 2011 22:07 UTC (Sun) by nix (subscriber, #2304) [Link]

That in part is why Bugzilla is an abomination unto FOSS (go through fifty steps, including account creation and verification, to file "The about dialog spelled 'GNOEM' wrong").

Oh look, it's a quote for next week's LWN :)

Who is that code for?

Posted Jul 4, 2011 1:17 UTC (Mon) by jrn (subscriber, #64214) [Link] (1 responses)

> Requiring users to create test cases because the developers are too lazy to do their damn job is also a problem.

Am I the only one who finds developers saying I have no clue whats causing that; heres some information that might be useful but your best bet is to come up with a reproduction recipe so I can investigate it on my end to be comfortingly honest?

Who is that code for?

Posted Jul 5, 2011 20:26 UTC (Tue) by elanthis (guest, #6227) [Link]

As a very last resort? Sure.

A dev can't fix a problem they can't reproduce. A dev also can't fix a problem they don't even try to reproduce, though. :)

Who is that code for?

Posted Jul 7, 2011 20:43 UTC (Thu) by ovitters (guest, #27950) [Link]

1. Bugzilla is not meant for end users. It is a developer tool. End users can file bugs if they wish to do so, but it still remains a developer tool. A lot of effort is done to ensure end users still can use Bugzilla (bugsquad and so on).

2. It does not take 50 steps to file a bug on (GNOME) bugzilla

3. Support should be done by a support team. If developers provide support, awesome. If not: unfortunate, but too bad.

4. Calling things hobby and so on is a bit strange. If you require support, shop around for it. There are various options for support.

5. Not sure if you meant this, so just clarifying to be sure: GNOME is not commercial, not run by commercial organisations, etc.

I'm an inactive Bugzilla developer and GNOME bugmaster btw.

Re: who should do the testing for the bugs?

Posted Jul 8, 2011 21:00 UTC (Fri) by oak (guest, #2786) [Link]

> Especially when it comes to end-user software, it's valuable to move the effort onto the developers, not the end users. [...] Requiring users to create test cases because the developers are too lazy to do their damn job is also a problem.

Valuable how and to whom?

If some software has, say thousand users per one developer, how it's valuable for developer to spend all his/her time trying to reproduce those thousand users' potential issues, instead of actually developing the software, improving its test-suite and fixing real bugs?

If user doesn't have a test-case, how developer is able to know that s/he managed to reproduce the "right" issue? And that his/her code change actually fixed the user's issue?

Most likely the user doesn't anymore even respond when developer finally has time to look at the bug. -> Bugs (or complaints) without proper test-cases or e.g. crash information that directly points out the problem, are just waste of everybody's time.

Who is that code for?

Posted Jul 3, 2011 17:47 UTC (Sun) by paulj (subscriber, #341) [Link]

Out of tree kernel modules are fairly rare though. There are a small number that get a decent amount of use. With those there's typically a commercial relationship (and hence responsibility, or at least pressure) for the provider of that out-of-tree (and usually closed source) module to support those users who decide they need it. Further, a big motivation with the kernel discouraging out-of-tree was about not having source - the kernel folks are usually *happy* to integrate and maintain modules to some degree.

With GNOME, it's a bit different. The GNOME shell developers aren't interested in integrating configuration options, rather they've designed the desktop so as to remove options from the core code-base, and instead push responsibility for end-user customisation to 3rd parties. Many of those 3rd parties will be hacker-users who miss some removed functionality, but who might not be in much of a position to give much support. In so far as any of these extensions become important to a significant number of users, the only place users will get support for them will be via distros. The distro engineers who end up working on these things sometimes will also be GNOME shell developers, quite likely.

I.e., from an end-user perspective, I see the 2 as being quite different. The kernel one exists both to ensure I pester my $PROPRIETARY_DRIVER vendor for support, and also to put pressure on them to consider submitting the driver for integration. Ultimately that works toward having better, less fractured support.

The GNOME shell one OTOH seems to fracture the support users can expect. If certain extensions might eventually become blessed and part of the core, that might be good. Till then though, it does seem more just a way to avoid responsibility for customisation options...

But we'll see in time, I guess. ;)