Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
Posted May 19, 2012 4:47 UTC (Sat) by geuder (subscriber, #62854)In reply to: Fedora 17 release pushed back to May 29 by rahulsundaram
Parent article: Fedora 17 release pushed back to May 29
From the hackers private point of view: little difference. The hacker gets 2 Ubuntu releases every year, he doesn't care about the presence or absence of LTS letters. The hacker gets 2 Fedora releases a year. My feeling is the Fedora is often somewhat more *bleeding* edge than Ubuntu. The hacker never sees RHEL, so he has has no idea what it is like.
From the enterprise point if view: RHEL somewhat irregularly every 2-3 years. Ubuntu LTS regularly every 2 years. Probably RHEL could be slightly more trusted to be stable at release time than Ubuntu LTS. Ubuntu will still fix afer release date if necessary.
From a vendor point of view: sounds like less work for Ubuntu than RedHat. because Fedora & Redhat seems really like 5 releases in 2 years (well, if 2 years were the target, but then they slip RHEL 5 stayed for 3 years)
From the hacker at work point of view: I clearly prefer Ubuntu LTS because universe repo seems 20 times bigger than EPEL repo. I sometimes have the feeling that there is even more ancient stuff in current RHEL than in current LTS, but I have not made any systematic statistics.
So overall it looks to me that the Ubuntu model works slightly better, but still Redhat seems to make more money.
Posted May 19, 2012 6:43 UTC (Sat)
by dlang (guest, #313)
[Link]
so this works out to even more releases
Posted May 19, 2012 6:44 UTC (Sat)
by carlm (guest, #84710)
[Link] (20 responses)
They are fixing on 3 years from now onwards.
> Probably RHEL could be slightly more trusted to be stable
RHEL is a helluva lot more trusted. Longer development cycle. Long history of being trusted. Larger number of paid developers, many of whom are authors/contributors to upstream projects. (As an aside, it's a real shame to see so little overall upstream contributions from Canonical compared to many other groups of developers).
> From the hacker at work point of view
Odd comparison, because that's not the target market of RHEL.
> but still Redhat seems to make more money.
That's because it's a helluva lot more trusted by the enterprise :)
Posted May 19, 2012 8:32 UTC (Sat)
by geuder (subscriber, #62854)
[Link] (17 responses)
Yes, by those who understand little of the daily work. (CIOs and the like)
But what do you when you need some software not in the RHEL repos? Install an rpm that Google finds on some random site? Start to build from source and need more dependencies which are not in the repo either...
I'm not convinced that such excercises typically lead to more trustworthy results.
>> From the hacker at work point of view
> Odd comparison, because that's not the target market of RHEL.
Why not? Because there are no hackers with the need to earn their living in a place using RHEL? Or because places using RHEL should not employ hackers???
Posted May 19, 2012 10:45 UTC (Sat)
by carlm (guest, #84710)
[Link] (1 responses)
Trust me, RHEL is considered the gold standard Linux distribution in the enterprise, for both people that understand daily work and people that don't. Carry on this debate once you've talked to 20+ large companies/establishments (hint: I have lots of experience in IT :-)
> But what do you when you need some software not in the RHEL
A stable base is *far* more important than niche software. Most important and regularly used software will already be packaged for RHEL/EPEL.
Regardless of which is better for hackers, RHEL is the clear winner for enterprise. Clearly the enterprise is the larger market, so don't be surprised that RHEL are earning more money :-)
Also, development work can easily be done on Fedora instead. And yes I've done development on RHEL too and it doesn't take that much knowledge to "make install" or create an RPM.
Posted May 21, 2012 11:42 UTC (Mon)
by dgm (subscriber, #49227)
[Link]
But WHY? I cannot claim your experience, but for the projects I work in, RHEL has been considered for only one reason: long support time. The projects I do usually spend 2-5 years _in development_ and would be in production for at least 10 more years. Nobody else can offer that kind of support.
If Canonical would have offered such support we would have considered it, because in my work place there are several people running Ubuntu, while exactly ZERO using RHEL or Fedora, and just one guy using SuSE.
Posted May 19, 2012 11:57 UTC (Sat)
by drag (guest, #31333)
[Link] (14 responses)
That's a load of crap. There simply isn't anything else to say about it.
> But what do you when you need some software not in the RHEL repos? Install an rpm that Google finds on some random site? Start to build from source and need more dependencies which are not in the repo either...
Repositories are _VERY_ overrated. Especially if you are doing enterprisy stuff. The reality is that you will need specific versions of specific applications and the chance that distributions actually support what you need out of the box is slim to none.
Business needs dictate application requirements. Besides the obvious corporate requirements; considerations include time tables, supportability, legacy requirements, available man power and talent, existing infrastructure, and interopability with other existing systems and such things. No two organizations have the same requirements. Therefore it is impossible for a company like Canonical to anticipate the needs of it's customers by using Debian repositories.
A enterprise server operating system is there to serve as a platform for your applications. It's nothing more then a concrete slab in this case. You want to put very minimal changes into it once it's deployed. The "Application Stack" you use on it is going to be something that meets your specific needs and specific goals. Once you are dealing with thousands of servers, dozens of different development goups, dozens of vendors, and thousands of customers... with each customer having their own requirements.. the value of repositories is completely nullified.
Imagine you must make data connections with customers. Each customer has their own needs and will require different versions of the same protocol. Maybe one customer wants Version 2 of the protocol. Maybe another wants version 6. Other ones will want binary streams while others want everything to be using the version that is wrapped in XML. This is not even a question of bad design or not... this is 'the customer is always right' and your business makes it's money by helping it's customers saving theirs. Part of that is being sensitive to their specific requirements and catering to them individually. You may face situations were you need to run a half a dozen different versions of the same software on the same server. 'Apt-get' isn't something that is going to be helpful.
Now Ubuntu LTS is fine. You are sure as shit not going to run a normal Ubuntu or Fedora release, but LTS would not be a terrible choice.
Redhat is popular because Redhat has done a good job in the past and people have depended on it. Enterprise Linux people don't use Redhat out of ignorance.
Posted May 19, 2012 14:26 UTC (Sat)
by carlm (guest, #84710)
[Link] (7 responses)
Posted May 19, 2012 20:51 UTC (Sat)
by dlang (guest, #313)
[Link] (6 responses)
there are many parts of the post that I agree with ( Fedora is a bad choice, the distro is the platform for your application)
But I do think that a lot of people use RHEL without thinking, and then limit themselves to only run what is in the RHEL repository, and this is far from the best way to do things.
Posted May 19, 2012 22:16 UTC (Sat)
by carlm (guest, #84710)
[Link] (5 responses)
Actually, by settled I mean I'm happy with the way the debate has gone and feel that readers of the debate will be less inclined to side with geudel's pov.
I wasn't dismissive as I tried to refute geudels arguments.
I understand your point but the way some people use RHEL has little to do with how RHEL and Ubuntu LTS compare with each other, which is what I thought this debate was about. If you want to take it in that direction, then clearly there are also people using Ubuntu LTS that don't have a clue what they're doing ;-)
Posted May 20, 2012 1:07 UTC (Sun)
by dlang (guest, #313)
[Link] (4 responses)
Posted May 21, 2012 12:03 UTC (Mon)
by drag (guest, #31333)
[Link] (3 responses)
Who claimed that?
Posted May 21, 2012 14:51 UTC (Mon)
by pboddie (guest, #50784)
[Link] (2 responses)
So said "carlm". RHEL, following the long tradition of Red Hat support, is pretty good for environments where platform stability and reliability are required, but that's not 100% of "the enterprise" by any means, and it's probably not 100% of the datacentre, either. I agree with those who feel that RHEL might be shoved into places it doesn't belong: that's where it becomes the "fool's gold standard". Actually, the "fool's gold standard" is probably Fedora when it is chosen solely on the dubious basis that some RHEL systems may be targeted by some development effort or other at some point in the uncertain future, and when the competition would provide a better experience for people to actually get their jobs done.
Posted May 24, 2012 5:57 UTC (Thu)
by drag (guest, #31333)
[Link] (1 responses)
I think what was quoted is not the same as claiming people that used something else as unprofessional.
Redhat certainly is very popular and in many ways it is a defacto standard that other 'enterprisy' distributions tries to emulate closely. Systems like Suse and Unbreakable Linux. This is done for application compatibility, of course. I don't know if this translates to a 'gold standard', but I don't think he meant that somebody using other things is inherently unprofessional.
Posted May 26, 2012 15:43 UTC (Sat)
by carlm (guest, #84710)
[Link]
That was a misrepresentation of my position, also known as a straw man ;-)
Posted May 21, 2012 9:36 UTC (Mon)
by epa (subscriber, #39769)
[Link] (4 responses)
Further, it's not possible to install a package just for one user, in that user's home directory, so a lot of the usefulness of a multi-user system is lost. You can still do it if you build from source, but that's a lot of fiddling and you lose the benefits of automatic dependency management and automatic application of security fixes.
That's one reason why virtualization is so popular. You run a whole new system for each application because the packaging tools we currently have fail at the task of providing isolation between different users, with different software requirements, on the same system.
Posted May 21, 2012 15:02 UTC (Mon)
by pboddie (guest, #50784)
[Link] (3 responses)
But it would be nice if the tools were prefix-aware and if packages were prefix-insensitive.
Posted May 21, 2012 22:37 UTC (Mon)
by idupree (guest, #71169)
[Link] (2 responses)
(The enterprise wouldn't like how NixOS doesn't have lots of people behind it the way RHEL or Debian do.)
Posted May 25, 2012 8:50 UTC (Fri)
by epa (subscriber, #39769)
[Link] (1 responses)
Posted May 25, 2012 8:51 UTC (Fri)
by epa (subscriber, #39769)
[Link]
Posted May 21, 2012 19:41 UTC (Mon)
by jspaleta (subscriber, #50639)
[Link]
Look at how Canonical's juju/charm initiative is actively making easier to bypassing repository packaging and pull web service infrastructure elements directly from git and build cloud instances from source as a charm using community best practice.
-jef
Posted May 19, 2012 8:43 UTC (Sat)
by misc (subscriber, #73730)
[Link] (1 responses)
Canonical have shown they can innovate, by trying lots of thing ( like Unity, like trying to have Qt on the desktop, the whole RB/Banshee/RB dance, having U1 on couchdb, then on something else, trying to keep bazaar and bzr, etc, etc ), but while that's good for a experimental distribution ( and that's IMHO the heart of free software ), I would say that tend to make some people cautious, because you do not know if what you use will still be there for the next version, and if you can really invest everything on this. Simple example, Canonical was a partner of Eucalyptus, then they changed their mind, what for people who relied on it ?
Hence the good idea of RH, to separate RHEL from Fedora, so people on Fedora are free to innovate, to do what they want, and RHEL people do not suffer from it ( on a reputation point of view ).
On the other hand, having Ubuntu as a one size fit all distribution is a little bit counter productive ( but at least, they got it right by having 1 company named differently from the distribution, unlike Mandriva and Mandriva Linux ).
Another weakness of Canonical is the model they use for now as it make it difficult for them to monetize their success ( at least on the server ). Since everybody can do support on their distribution, sooner or later, someone will do it. In fact, it already happen, smaller company doing support without Canonical getting anything ( french assembly is the perfect example there ).
Again, Novell and RH likely did anticipate this by having 2 separate branches, and 2 different model for it. So they can have enough "exclusivity" on the entreprise side to have a sustainable model to fund the second side.
Instead, Canonical is trying to build exclusivity by using their CLA ( so someone wanting to use ubuntu on TV need to sign with Canonical or distribute all the code as AGPL v3, for example ), or by using the trademark ( like for the ubuntu one discussion ), both solutions being not that great for some part of the communities around free software. But they are right to try different solution, that's always nice to see people trying stuff.
Posted May 19, 2012 12:05 UTC (Sat)
by drag (guest, #31333)
[Link]
They didn't antipicate this so much as 'Learn this the hard way'. Redhat tried to do 'one release for everybody' and it was a disaster. They tried to equally support Gnome and KDE while papering over differences, and turns out it is a terrible and stupid thing to do.
It took them years to figure out that they needed to have something like Fedora.
And, at least in Redhat's case, exclusivity excuse is a complete red herring. Almost everything they do is open source. Linux OS vendors that tried the 'proprietary value addition' approach are all gone and buried because it's a shit approach to retaining customers.
Posted May 24, 2012 19:56 UTC (Thu)
by trulyexcitingnickname-dontuthink (guest, #84181)
[Link] (1 responses)
except that Ubuntu's LTS universe is out-of-date and buggy versions of Debian packages...
Posted May 24, 2012 20:03 UTC (Thu)
by Cyberax (✭ supporter ✭, #52523)
[Link]
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
http://www.pcworld.com/businesscenter/article/255629/red_...
> at release time than Ubuntu LTS
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
> Yes, by those who understand little of the daily work.
> repos? I'm not convinced that such excercises typically lead
> to more trustworthy results.
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
Trust me, RHEL is considered the gold standard Linux distribution in the enterprise
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
You may face situations were you need to run a half a dozen different versions of the same software on the same server.
This is the weak point with current packaging systems like rpm and dpkg. An application is installed, centrally, with a single version, or it isn't installed. You can have multiple versions of libraries but (on the rpm side at least) this requires you to make special 'compat' packages.
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
To answer my own question:
Fedora 17 release pushed back to May 29
Starting at version 0.11, Nix has multi-user support. This means that non-privileged users can securely install software. Each user can have a different profile, a set of packages in the Nix store that appear in the user’s PATH. If a user installs a package that another user has already installed previously, the package won’t be built or downloaded a second time. At the same time, it is not possible for one user to inject a Trojan horse into a package that might be used by another user.
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
- demonstrated they are here to stay ( ie, by having a sustainable business model, something that Canonical still seems to search, unfortunately for the whole free software ecosystem who would surely be stronger with one more player )
- showed they were conservative enough, and that's one of the other weak point IMHO of Canonical.
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29
Fedora 17 release pushed back to May 29