Submitted by Mozilla’s Firefox browser will probably abandon support for older distributions of Linux, a company executive has said. Mike Conner, director of Firefox development, has proposed that the browser’s next edition work only with more-or-less current Linux runtimes, rather than support aged versions.
Firefox May Scuttle Support for Older Linux
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
What is their definition of an older distro? And does this mean that Firefox is going to evolve into a mess of dependencies like other software?
(Quite frankly, I like the ability to download the up-to-date binary from their website and have it installed in a jiffy, regardless of my distribution.)
Their definition of “old”, according to the german Linux news site pro-linux.de, is Ubuntu Dapper Drake (last year) and anything older. Quite hefty, if you ask me, because this will also exclude many enterprise oriented, more conservative distros.
Well reading his blog (from the story), here are the proposed new requirements:
http://wiki.mozilla.org/Linux/Runtime_Requirements
It doesn’t look to me bleeding edge at all.
So post-Dapper/Etch is not bleeding edge?
If not, I don’t what is.
Well reading his blog (from the story), here are the proposed new requirements:
http://wiki.mozilla.org/Linux/Runtime_Requirements
It doesn’t look to me bleeding edge at all.
Yeah it cuts off Dapper thats not reasonable.
Basically Mozilla ships precompiled firefox builds built using GCC 3.x along with glibc 2.3.2. That box runs Redhat 8/9 I believe.
To run these builds on any recently linux distribution you have to install the comatibility library for GCC 3 called libstdc++5 or similar on most distros as most have moved over to GCC 4.x now.
They are using this as their “target” platform and all code had to build successfully on this machine. So I guess now they are planning of moving to newer build boxes possibly with GCC 4.x and so wont have to hack around their code so it perfectly compiles and runs with the now legacy GCC 3.x and glibc 2.3.x on their current build systems.
Oh and of course they are also only depending on older versions of libraries such as GTK2, pango etc so they have to hack their code to make sure it works with these older libraries.
But as they mention of course since its open source you will find patches that will still enable you to compile with legacy systems if required.
Edited 2007-05-16 21:34
It helps innovations. It leaves less time for vulnerabilities explotations. It leads to cleaner and more robust code.
> It helps innovations. It leaves less time for
> vulnerabilities explotations. It leads to cleaner and
> more robust code.
And it is a pain in the arse to keep with, especially if youre a unknowledgeable desktop user, or working for a company which doesn’t update their distribution every single year.
It wouldnt be a bad idea for them to make a release policy like Ubuntu: provide LTS releases every two years, and keep them up to date for four or five years, regardles how long you support the smaller releases in between.
Cutting off everybody except the last few bleeding edge releases is bad, just bad for the perception of Linux.
You get what you pay for.
You get what you pay for.
You disagree?
Yup, youre right. Im 100% positive that any second now people all over the world will see the wisdom you presented in your post and start going back to IE in droves.
You’re implying that IE8 will be backwards-compatible with Windows Vanilla XP. Will it?
They don’t say it is not possible to backport. They say that *they* aren’t doing it.
Even if they backported to Ubuntu LTS, Ubuntu would most likely not include it in their repositories; versions are mostly frozen upon release, and only security updates or fixes for serious flaws are pushed. Debian would likely do the same.
As for Redhat or SUSE enterprise distros, they are free to backport 3.0 if they want to. For the money they charge, I’d expect them too.
Anyway, if you like using old distros, and don’t want to upgrade, then you can bask in the warm feeling of enjoying old friendships, and just keep on using FF2; it is not as though it will stop working just now.
I agree… If you are using old distros – then what’s the point in getting the latest Firefox anyways? Debian Sarge only had Firefox 1.5 on stable for the longest time. So people with old distros can just stay with their old versions of firefox (like 2.0.0.3).
When there are showshopper bugs or when the hardware retires – then it’s time for them to move up to 3.0.
People are making this a bigger deal as it is.
Individuals staying with old distros are probably not going to use bleeding edge releases anyways.
Actually Debian Sarge has Firefox 1.0 + lots of patches by Debian (unless you use backports). I still have Sarge on a couple of machines running. I haven’t found time to switch to Etch yet. Maybe when Debian 4.0r1 is released…
“””
Anyway, if you like using old distros, and don’t want to upgrade,
“””
Except that some of these dependencies are only a few months old. About the same age as Windows Vista. Perhaps they should cut support for XP, too?
Debian caught a lot of flak when they decided to rename FireFox for Etch. Mozilla has shown less interest in linux and Mac than Windows (b/c that is where most of their users are). Because of this, getting Mozilla to approve changes is a slow process. Now, they might stop approving all changes for older (1.5.X) versions. Or, I think they would take so long in approving those changes as an incentive for distributions to move to the 2.X series of FireFox. Enterprise distributions and their customers value stability, and if Mozilla is unwilling to help support older versions of Firefox then more and more distributions might use Firefox under a different name. There is really nothing wrong with this. Collaboration between major distributions will keep the 1.5.X browser secure and viable in the enterprise workplace. However, I wish they would find a more suitable name than Ice Weasel and Ice Ape.
Do you have anything to backup your claims about Mozilla showing less interest in Linux? If you keep in touch with Mozilla development you’ll see that if anything the OS X version probably gets the least attention.
At any rate it is irrelevant discuss maintenance/long term support issues without looking at the specific of every platform. By its nature Linux and the ecosystem surrounding it does not emphasize backwards compatibility, instead it tends to stay pretty close to bleeding edge. It’s one of the benefits of open source that you can patch it yourself if Mozilla doesn’t want support it. And it’s not like Mozilla is being uncooperative on this issue. They accept contributed builds as long as your build is up to standard.
And it’s not like they’re dropping legacy support for only on Linux, pre-win2000 systems and OS X 10.2 are also getting the axe. They have to move on, if they had the resources to support these systems, I am sure they would.
For most enterprise distributors this is a none issue. Red Hat have an awesome relationship with Mozilla and I don’t think Red Hat are as paranoid as you are. And people who run Debian will simply use Debian’s patched builds.
Your post reminds me of a typical FOSS camp reaction to decisions made my Mozilla that are perceived to violate the FOSS ethos. Remember the creation of the Mozilla Corporation? So has anything changed?
The further back you have to support something the less resources can be assigned to improving newer versions. I suppose that’s the game software developers have to play, but people who continue to use older software must realize they are dead weight. I don’t mean that in a harsh way. Many old programs still do the job just fine and I see no reason to toss older hardware or software if it still performs the job it needs to. However, at some point software updates and new programs aren’t going to be available.
With all that said I don’t know how far back Mozilla is talking about. I would assume they are talking about a few year from the release of version 3, but I’ve been wrong before.
Imagine if Microsoft tried to do the same thing…
Like making IE7 only run on XP (leaving out 2K) and up?
Yep. The response was quite different.
Or making Direct X 10 only run on Vista?
If it weren’t for Firefox, XP wouldn’t have gotten IE7.
If it weren’t for Firefox, nobody would have got IE7.
“Like making IE7 only run on XP (leaving out 2K) and up?”
Yes, like Microsoft doing that, and thanks for proving the sappyvcv’s point. Because the fact is that Microsoft was trashed on message boards for doing that. Yet Mozilla doesn’t get trashed for their plans to drop Win9x and old Linux distro support for Firefox. On the contrary, I see a lot of Firefox apologists spinning this away. And Apple didn’t get trashed for making Safari 2.x require OSX Tiger.
Edited 2007-05-17 13:59
From the Mozilla wiki:
Proposed Runtime Requirements
* GTK+ 2.10.x
* GLib 2.12.x
* GNOME 2.16.x
* Pango 1.14.x
* Cairo 1.4.x (only for system cairo)
* xorg (libX11) 1.0.x
* dbus 1.0.x
* hal 0.5.8
* libjpeg v6b
* libpng 1.2.x (tracking upstream)
* zlib 1.2.3
Wow, to me this does seem pretty bleeding edge. What about operating systems like OpenBSD that has gnome 2.10 and does not use hal or dbus? What about Debian Etch for that matter that ships with Gnome 2.14? Why does a web browser need dbus or hal anyway?
Why would you use OpenBSD as a desktop?
Nothing seems to be stopping distro’s from backporting bug fixes etc… so I don’t see the issue.
It might lead to more distro’s forking Firefox though, as Debian has already done.
> Why would you use OpenBSD as a desktop?
Because you can. I use it here w/o problems and the system is so rock solid that i won’t remove it for nothing.
/*Why would you use OpenBSD as a desktop? */
I use OpenBSD everyday as one of my desktop OS. I build websites, practice my progamming,surf the net,etc on OpenBSD it does everything i need as a desktop OS.just because it does not have little cute configuration menus,wizards like many OS it does not mean it can not be used as a desktop OS. I feel i need to post this from OpenBSD fag site:
1.10 – Can I use OpenBSD as a desktop system?
This question is often asked in exactly this manner — with no explanation of what the asker means by “desktop”. The only person who can answer that question is you, as it depends on what your needs and expectations are.
While OpenBSD has a great reputation as a “server” operating system, it can be and is used on the desktop. Many “desktop” applications are available through packages and ports. As with all operating system decisions, the question is: can it do the job you desire in the way you wish? You must answer this question for yourself.
It might be worth noting that a large amount of OpenBSD development is done on laptops.
“””
I feel i need to post this from OpenBSD fag site
“””
I happen to be Gay, myself. And I’ve been a Unix/Linux admin for 19 years.
But I have never heard of this “OpenBSD fag site” you refer to. 😉
/*But I have never heard of this “OpenBSD fag site” you refer to. ;-)*/
sorry, it’s a typo, i meant faq as frequently ask questions.
http://www.openbsd.org/faq/faq1.html#Desktop
Edited 2007-05-16 22:28
“””
sorry, it’s a typo, i meant faq as frequently ask questions.
“””
Well, I kinda figured. 😉
“Why would you use OpenBSD as a desktop?”
Because it suits your needs.
Why use OpenBSD as a desktop? Well, why not? It is stable, secure, well documented nice little (small and simple is good!) operating system. The only problem I have encountered is that the OpenBSD packages/ports collection is a little bit limited but I’ve found that all the “essentials” are there (yes, soon even OpenOffice port will be ready, with a huge amount of bugfixes).
OpenBSD offers lots of security features which are up and running by default (you might want to see slides on “Exploit mitigation techniques” http://www.openbsd.org/papers/ven05-deraadt/index.html). Many of these techniques help securing things such as web browsers and other desktop apps (fully randomized malloc(3) for example).
One might also choose OpenBSD because they offer fixed release schedule (new release is made every 6 years, a release is supported for one year). This helps to plan upgrades. Installing new versions is also usually very painless because new versions are evolutionary improvements over the old versions (as opposed to revolutionary, usually revolutionary means “causes breakage and nightmares to sys admin”).
FYI, OpenBSD is already porting gnome 2.18.1. Doesn’t have hal though, but dbus is already on it.
alemao:4$ pgrep dbus ; uname -a
11086
2045
2659
OpenBSD dub.int 4.1 THC#0 i386
OpenBSD running gnome 2.18.1:
http://humppa.nl/~jasper/gnome/screenshots/gnome5.png
Edited 2007-05-16 20:57
Pardon? 2.16 was released last year, 2.18 has been released in the last couple of months, and by the time 3.0 of firefox is released, 2.20 will almost be complete – three whole release cycles of GNOME have come and gone – I’d hardly call that ‘bleeding edge’.
With that being said, I don’t know where HAL falls into it, I could possibly understand dbus, but in the same situation, I doubt it would need 2.16, more like certain components from GNOME 2.16 such as ORBIT and the likes.
“What about operating systems like OpenBSD that has gnome 2.10 and does not use hal or dbus?”
OpenBSD has dbus (but not HAL) and gnome 2.18 is in the works. Also note that it says *Linux* requirements. OpenBSD isn’t Linux.
Edited 2007-05-17 04:31
because dbus is a great tool for interprocess communication, so any program that could make use of interprocess communication should use it.
Better questions are:
Why does a web browser need a text editor with spelling correction built-in?
Every linux distro includes many text editors all of which are better, why not just cause them.
Why do I need, CSS and javascript support to view 80% of pages I visit?
Because designers broke the web and nobody will be able to fix it.
because dbus is a great tool for interprocess communication, so any program that could make use of interprocess communication should use it.
Sure, if you don’t want to be portable. That’s fine if you don’t want you code to be portable, but then it isn’t fair to claim your code is portable “because you can port my dependencies, too!” which is pretty much what Mozilla does.
Since you are replying to a statement about D-Bus: I think the only Mozilla target platform which doesn’t have a fully implemented D-Bus yet is Windows, where the developers still need to (AFAIK) sort out some of the things that are different on Unix/POSIX, e.g. user credentials, transport (instead of unix domain sockets)
Currently, sure. What about other platforms? BeOS doesn’t use dbus. Nor does SkyOS. If we want to port Mozilla to Syllable, we have to port dbus (Even though we already have a perfectly good IPC mechanism).
As you point out yourself, it’s difficult to even get it to work on Windows, where there are far more developers looking at the code. New and innovative systems are burdened with a whole bunch of dependencies which they don’t want and shouldn’t need. dbus is a system component of a certain operating system and it was designed to solve a problem on that platform. Same for HAL. Dragging them along to other platforms is just added cruft and extra work.
In fact there are more developers working on the Unix ports than on the Windows side. IIRC there are only two or maybe three KDE/Win32 developers working on D-Bus/Win32
Well, it isn’t a component of an operation system or developer for an operating system, but rather a component used in and developed for free software desktop applications.
Sure there are other technologies available for offering services to other applications, e.g. CORBA, but the experience with DCOP has shown that a simpler one, which is even usable from commandline/scripts, gets incorporated a lot faster and is useful not only for other application developers but for users as well.
My point was that there are more developers for the Windows port of dbus than a small operating system could possibly hope to have. Two developers for the Win32 port may be less than the Unix version, but it is far more than an OS such as Syllable could hope to have.
Well, it isn’t a component of an operation system or developer for an operating system, but rather a component used in and developed for free software desktop applications.
That’s fine. But it’s unfair to characterize it as “the” solution to everyones problems. DCOP or CORBA or other schemes may be “worse” than dbus, but any given OS may have a “better” solution, but will be forced to port dbus anyway because it is a dependency of the Mozilla codebase.
I don’t have a problem with dbus itself. My point is that rather than try to ensure the codebase is truly portable, the Mozilla developers have simply decried that they will rely on a certain set of components that must also be available on any given platform. That makes their lives a little easier, but it complicates matters for anyone who wants to port Mozilla to a new platform. It is an extension of the “All the world is Linux”[1] mentality that exists throughout a large segment of the Open Source world.
[1]: Older developers may remember “All the world’s a VAX”. Same problem, different context.
For example detection of changes on network hardware, e.g. recognizing offline states, switching to a different caching policy when on modem, etc.
Or different animation/plugin policy on low battery.
Just because an application does not use any hardware directly doesn’t imply that it can’t improve the user experience by knowing what is going on.
I see a proliferation of not-quite-but-almost-Firefox forks in the near future.
Agreed…
Or more adoption of Epiphany and similar…
Makes Konqueror much more appealing too… especially if they do import WebCore in KDE 4.x
I wonder if the Firefox maintainers realize they’re killing their browser right now?
Makes Konqueror much more appealing too… especially if they do import WebCore in KDE 4.x
I’ve been working on a fork of konqueror to focus more on the web browser usability. Nothing really worth releasing at this point, but I’ve gotten a pretty good look at the state of khtml in kde 3 and 4 as well as the current status of the qt webcore port. Webcore’s still got a fair walk ahead of it before it’s really usable outside of osx. Obviously it’s rather nice though, even if a bit rough around the edges at the moment. If I recall, the kpart hasn’t been updated in some time now and I don’t think it even builds anymore. Still, I didn’t give it too much of a try as I was more interested in just getting the gist of how well the qt port in general was going.
Saying that though, I’ve been really surprised by how nice khtml has become. I always thought of it as little more than a toss in for reading help files in html, but it’s really pretty darn nice. The updates for kde4 even more so…especially once the ssl code is updated. It’s really easy to write a browser around the rendering engine as well. I’d never really done any coding with kde before, and doing so has impressed me pretty highly. After I initially became disillusioned with firefox’s performance in linux, or at least on my linux install, I played around a bit with its code. For the most part I found it fairly painful. It just wasn’t very fun code to work with. konq/khtml on the other hand is actually ‘enjoyable’.
If anyone ever felt like building a browser around khtml and the related libs, they’d have something comparable to firefox in hardly any time at all. I had almost no free time, no experience at all with kde coding, and wound up implementing what I thought would be fairly huge inclusions in about a weeks time. I have little doubt that someone more experienced could have build a browser around the bare khtml/kjs/etc in a couple days.
I hope your project leaves to see the light of the day outside your development machine and then manage to get some contributors and shape into something good enough to compete with Firefox, but based on KHTML/WebCore instead.
While I have nothing against Dolphin per se, I think that this time the KDE developers took the wrong approach. There wasn´t anything wrong with Konqueror as a file manager (although I do agree that the UI could have be made a little bit simpler). Quite frankly, it is the most powerful (arguably the best) file manager in existance right now. It was the web browser portion of it that required some care. Even using different profiles, things like clicking on the Home icon would lead you to to $HOME on your file system instead of the home page on the web browser. The configuration window can be rather confusing at times, mixing file manager and web browser options in the same panels.
No, they got it wrong. Instead of working on Dolphin, they should have ditched the Konq-as-a-web-browser idea (Konq should keep its ability to use KHTML as a KPart, though), get KHTML back in shape with the changes that Apple made in WebCore and create something along the lines of Safari. That would have been much better.
But I will reserve my judgement until KDE 4.0 is released in a usable form and give some credit to the devs for the time being, as I have yet to be disappointed by these clever guys.
I don’t. This whole announcement is about reducing the amount of Linux only hacks in the codebase, that increases maintenance and disro-specific patches, the cause of many Firefox-on-Linux complaints.
There are still plenty of distro’s that are not willing to require users upgrade… and those distro’s will simply fork Firefox, and maintain their own code.
I believe Ubuntu is even following Debian already, so that already accounts for more than half of current Distro’s.
It will happen, although I commend Firefox for trying to simplify the codebase. Distro’s like Fedora will be fine with it I think.
Why? Perhaps there´ll be one or another more unofficial build of firefox but I can see no reason why anyone would fork firefox for this reason.
/* The open-source browser currently accounts for 15.4 percent of the global browser market.*/
I thought the firefox browser had a larger market share. I’m sure when firefox 3.0 comes out all the linux distros will be up to date to run it without any problems.
Depends what market you look at. On weekends, some countries have over 30% usage. Most people don’t have control over their enterprise desktops, so don’t have a choice in what browser they use.
/*Depends what market you look at. On weekends, some countries have over 30% usage. Most people don’t have control over their enterprise desktops, so don’t have a choice in what browser they use.*/
understandable opinion on why firefox has a global 15% market share.
Seeing how they HAL will be required, how will this work with OSes that dont have a Linux HAL implimentation?
“Seeing how they HAL will be required, how will this work with OSes that dont have a Linux HAL implimentation?”
Those usually aren’t Linux.
It’ll probably be configurable (aka ./configure)..
If not, I’ll bi*ch slap each developer..
OpenBSD makes a great Workstation, Using it as we speak to make this post..
I love firefox news is *always* never about firefox.
Relatively modest hardware will run a modern distribution, and by modest I mean years old. The problem with the main *linux* Desktops is they in the Linux world have needed lots of memory, but if you could run them then, you can still run them now. It is not like you are *paying* for proprietary binary only applications that rely on either *backward compatibility*. You can simply update *everything*, and the advances in the past 6 months are always worth the upgrade.
I noticed that the first post was about downloading an application from the website to work on all distributions. I think that is wrong, but highlights *again* that *cutting edge* distributions should work with or around the Mozilla team to ensure that delivery of popular applications are available swiftly, and bug free.
I also noticed someone had been modding down for suggesting forked projects. I’m surprised at this especially as it seems such an obvious thing. I prefer firefox 2.0 I could never go back to 1.5 or even earlier, but It isn’t as stable and does *feel* slower. Admittedly plug-ins take care of a lot of functionality that otherwise could have caused a folk, but Seamonkey has shown that plug-ins do not take care of fundemental things. The gaps between releases of firefox have gotten larger as they have got more ambitious. Admittedly I would not like to go back to the continual releases of the old Netscape days that added nothing new. Open source products work better with a faster releases, and smaller incremental changes.
On a final note OpenBSD/Microsoft users what on earth is going on there. I cannot see the relevance to BSD users as simply the license does not allow you to use changes to the base code without releasing the source. Perhaps you should stick to mentioning Swiftfox, and why it wont release the source? Or Microsoft users on topic why are you not mentioning what the next version of Firefox won’t run on pre-XP OS’s even thought their are a possibly!? more users running those than there are linux users.
now that older distros are not supported, someone can go and fork the new Firefox, trimming the newer dependencies…
Of cource, this assumes the common misconception in regard to Open Source that:
1) There are enough people with the skills to do it, other than its’ current programmers.
2) That they will bother doing it.
People having an itch plays no role. There have to be developers having an itch, and skilled ones in the particular problem domain at that.
It makes sense since legacy support is a bitch.
A lot of the dependencies are because of the changes in X in the last few years. Some of the things in the current Linux packages just cannot be supported by older X versions. And on top of the newer X are the newer Gnome and KDE libs that take advantage of it. So the “bleeding edge” is not arbitrary. There are good reasons for it. For example, would you like to regress back to the ugly non-AA fonts of only a couple of years ago?
There are more than enough free or open source webbrowsers out there (some even using the same mozilla engine) so if firefox becomes too ‘cuttong edge’ then just switch to one of the number of other capable browsers out there.
Besides, it’s not like most people /need/ half of the extra features firefox ships with
Edited 2007-05-17 09:20
I have a bit of a problem with this. I usually stay about two or three versions of linux behind the regular version. I currently have FC3 and FC5 on my home systems as well as Centos 5 on my server at home. While I have no problems with FC5 on new hardware, it doesn’t seem to like older hardware. I kept a copy of FC3 for an Athlon processor and upgraded the software using source code builds.
On both my FC3 and FC5 systems I have the same gnome and same Firefox, and others as well. If Mozilla is going to drop binary support for older Linuxes that’s acceptible. However if they are going to make their source code incompatible with old versions of Linux then I see this is as a big problem.
But you’ll still be able to use the same version of Firefox that you have been using. Only new releases will not be backwards-compatible out of the box. I’m sure that bug patches will continue to be released for the 2.x series of FF for a while yet.
If they begin demanding gnome for running firefox on linux, im not even gonna bother have firefox installed. I use konqueror anyway..
OMG, say it ain’t so!