Please see Bug 577290 for more information.
Platform requirements here: https://wiki.mozilla.org/Firefox/4/Platforms I'll need some help coming up with: - Linux system requirements (minimum library support, etc) - Hardware requirements for reference, these are our current specs: http://www.mozilla.com/en-US/firefox/system-requirements.html
For Linux: We could recommend Xorg server 1.7 or higher for optimal functionality (bilinear filters for image scaling) and performance. We should still run with the same GTK+, GLib, Pango versions listed previously as we build (and still run some basic tests I think) against those versions. We don't run 32-bit plugins in 64-bit browsers, but 64-bit plugin in 64-bit browser architecture is the same as 32/32.
We require that the target Linux system come with libstdc++ for GCC 4.3 or newer.
We would like this to hardblock the release, because setting the right expectations will prevent a number of people from being really upset. It might not block Fx4 as piece of software, but this should be in place before we launch.
Created attachment 505623 [details] Example of columned view in HMTL It would be nice to show the three Operating Systems (Win, Mac, Lin) that are officially supported in columns, not listed, view - with a link to the page on downloading the source code underneath the three.
Is 32-bit builds only still correct?
Created attachment 506683 [details] FF4 SysRequirements (absolute links) This is a continuation of the previous example of SysRequirements in columns - except updated for FF4. NOTE: the links are all absolute and this should not just be copy-and-pasted onto the server. I will make a patch for the website later.
When working in the /projects/mozilla.com/trunk SVN tree, how do you manage the hundreds of language directories? Do I just work in en-US and ignore the others? Do I have to translate my additions for every language manually?
(In reply to comment #6) > Is 32-bit builds only still correct? We are shipping 64-bit releases of Firefox 4 on OS X and Linux. Note that the OS X builds are Universal 32/64-bit, so they should run on 10.6 or 10.5 on either x86 or x86-64 CPUs.
I would generally like this bug fixed before the 4.0 release... The Beta 11 Release notes are still linking to the 3.6 system requirements.. which I am pretty sure will not work with 4.0. On another note, this will be somewhat important in future troubleshooting.. as people will be wondering why Firefox will not run on their PPC Tiger machine.. or why Firefox runs slow on their machine with 64MB of RAM... I know it is rare that people will have such a small amount anymore... but I'm betting there are still a few out there. I would like a clear way to know if your video card supports hardware acceleration too.. but thats another matter.
Well Attachment #506683 [details] is an edit of the current 3.6 page. the important bit is the bit in <div id="main-content" />. That's a table of the requirements as documented here and in Bug #577290. If someone who can be bothered with CVS (why aren't we on git or Bazaar?!?) then they should basically copy that <div /> to the version-controlled page.
For a bit of perspective.. the minimum system requirements to install OSX 10.5 is here: http://support.apple.com/kb/HT3759 Summary: 867MHz or faster.. and 512MB of RAM Also.. For Windows Vista.. http://windows.microsoft.com/en-us/windows-vista/products/system-requirements Summary: 800MHz or faster.. and 1GB of RAM I know the needs of 2000 and XP may be less than Vista and Win 7.... though I don't know if there should be different requirements for different Windows versions... as that may get confusing.
All the requirements are there for is to say what we expect to work. Firefox comes with absolutely no warranty. So it's ok for them to be semi-simplified.
As bsmedberg said earlier, there is a minimum requirement on the Linux platform for libstdc++ 4.3.0 or greater when running Firefox 4. Whilst all the other Linux requirements in attachment 506683 [details] are exactly satisfied in CentOS 5.5 (implying that was the system used for establishing them), the libstdc++ requirement is not, as detailed in bug 621704 (the workaround for CentOS 5.5 is to extract a later libstdc++ library - e.g. from Fedora 9 - and put it in the Firefox install tree). Hence, I would suggest that the libstdc++ requirement is added to the Linux list - I have been running Firefox 4.0b9 and later betas with the aforementioned workaround fine on CentOS 5.5.
(In reply to comment #16) > It would be interesting for me if i can assume SSE2. I know 2 people who still run Pentium3 machines, one has a dual P3/866MHz machine and the other has a single 1.2GHz CPU. Both on WinXP. My observation is that Firefox 3.6.x runs pretty well on both machines. Changing the minimum requirements to a SSE2-capable processor will have a lot of people baying for your blood.
We made the decision several months ago not to require SSE2 for Firefox 4.
We have no automated test coverage for non-SSE2 machines.
Created attachment 515416 [details] FF4 SysRequirements (absolute links) Updated the requirements to reflect recent comments. Also tweaked some minor HTML A patch still needs to be formulated by someone.
Created attachment 515418 [details] <div id="main-content" /> to <!-- end div#main-content --> This contains only the main-content div (<div id="main-content" /> or div#main-content). Note, the XML comment <!-- end #content div --> signifying the end of the <div /> has been replaced with <!-- end div#main-content --> for clarity. To create a patch from this, simply replace everything from (and including) <div id="main-content"> to/and </div><!-- end #content div --> with the attached text.
Created attachment 515487 [details] [diff] [review] Patch with conforming markup Someone'd need to write some CSS, I suppose.
What's missing from the CSS, it looked fine on the "mockup" HTML?
(In reply to comment #19) > We have no automated test coverage for non-SSE2 machines. http://tinderbox.mozilla.org/showbuilds.cgi?tree=GeriatricMachines
From my understanding... the system requirements are based off of what the minimum hardware is needed to run the minimum supported operating system. ... If we keep the rule that it just needs the minimum needed for the OS... then I have some things to note that should be changed: 1. For Windows 2000, the minimum needed RAM is 32MB of RAM, (64 Recommended)and 133MHz processor. 2. For OSX 10.5. The minimum needed RAM is 512MB of RAM (Or More) And an Intel Processor (As Power PC's are no longer supported) ____ When it comes to computer games, a lot of times people look to the required and recommended hardware, to know what they need, in order for the game to be "usable"... I don't know if 32MB or even up to 512MB on a 2000 machine would be "usable"
32MB is not realistic. 128MB maybe.
comment #24: 4 columns are perma-red, 3 perma-orange. My point exactly.
Bug 579078 was where the decision was made on Windows 2000 support. I asked if tests were being run on it, but didn't get a reply (see bug 579078 comment 23).
(In reply to comment #23) > What's missing from the CSS, it looked fine on the "mockup" HTML? The "conforming" attachment does not match the earlier mockup at all. It has vertically centered table cells, the table headers are not sized, etc. In other words, <tr valign=top> became <tr>, <td...><h2>...</h2></td> became <th>, etc. In addition, both the mockup and the patch have no horizontal padding between columns. The columns are an improvement on the Firefox 3.6 requirements page. http://www.mozilla.com/en-US/firefox/system-requirements.html
The patch copy & paste the contents of Attachment #515418 [details] into the new page and update the title/etc. It should not edit Attachment #515418 [details] at all when pasting. The patch should also move the 3.6 page into firefox/3.6/system-requirements.html Also does anyone know why when set to width 100%, it is only the width of the header "System Requirements"? > In addition, both the mockup and the patch have no horizontal padding between > columns. Is padding really necessary in this case?
Attachment #515418 [details] has things like '<table width=100% boarder=0>' which is a typo. The "conforming" attachment removed the HTML attributes of <table>, <tr>, <td>. So it needs 'valign="top"' or equivalent CSS. The Attachment #515416 [details] mockup shows why horizontal padding between the three columns is needed.
... To further my quest... to have some realistic RAM requirements... I would like to also point to the Mozilla Thunderbird System Requirements... which seems to be way more power intensive... then FF4... http://www.mozillamessaging.com/en-US/thunderbird/system-requirements/ On Thunderbird... the ram is listed out per OS... and actually has somewhat near sane numbers.
Right now, you're talking about hardware requirements only. What about software ? Specifically, that 'HAL 0.5.8 or higher' part. As probably everyone knows by now, hal has been put in fixes-only mode awhile back and while its replacement (udisks/upower combo) has it's own shortcomings (my personal pet peeve is http://bugs.freedesktop.org/show_bug.cgi?id=25613), major desktop environments have moved to it either awhile ago or lately. AFAICT, firefox isn't even using it directly, just via gnome-vfs (which has been deprecated for even longer and 2.0 of xulrunner has a gio module to replace it anyway). (so OK, parts of what they do with Gnome 3.0 are more than a bit crazy, but that's a different topic) Perhaps a minor clarification should be added there ?
What about Windows Server 2008?
is this still not landing till the 4th when RC is planned to be complete before than?
Any strong reason why this is filed as a Firefox bug vs a mozilla.com one? If not, I'll move it over to that component so it can batched with the rest of the site release.
(In reply to comment #36) > Any strong reason why this is filed as a Firefox bug vs a mozilla.com one? If > not, I'll move it over to that component so it can batched with the rest of the > site release. I was wondering this myself because this really has nothing to do with firefox 4 itself more to do with the website not being fully up to date.
(In reply to comment #36) > Any strong reason why this is filed as a Firefox bug vs a mozilla.com one? If > not, I'll move it over to that component so it can batched with the rest of the > site release. It's a site issue, so it should be moved there.
(In reply to comment #36) > Any strong reason why this is filed as a Firefox bug vs a mozilla.com one? If > not, I'll move it over to that component so it can batched with the rest of the > site release. Likely because the Mozilla.com component doesn't have the same blocking-Firefox flags...
Slater, this bug is actually more about me collating and gathering the system requirements so we know what they are before we ship. Anyway, I'll have it done in an hour or so.
This is now the last blocker for FF4, according to http://canweshipyet.com/
there are now two blockers so this isnt the last one anymore..
Can we unblock RC on this?
This doesn't block RC, it blocks ship.
Sorry to ask this here, but since the can-we-ship-yet list was mentioned... Will RC/final ship for all languages and all systems (win64 is not available in quite a few languages for example), or will we ship with en-US for all systems initially and then start building for other systems/languages?
It will ship for all supported systems, and all locales. Win64 is not a supported build in FF4.
what will happen with FF4 x64 nightly build after ship, no more new x64 nightly?
File updated in r83983 on trunk; I'll go through and update all the links for release note pages before pushing live. The plan is: - retarget all 3.6.* links to system-requirements-v3.6.html - target 4.0 release notes to system-requirements-v4.html this will make system-requirements.html obsolete. Yay.
Can't we put it in en-US/firefox/4.0/system-requirements.html and en-US/firefox/3.6/system-requirements.html? Seems like that would be a bit cleaner (or even /system-requirements/index.html like details, whatsnew, and release notes)
Man, you should have seen this coming ;)
Is Firefox is compatible with 64-bit Windows?
Firefox will run just fine on 64-bit Windows as a 32 bit process.
(In reply to comment #53) > Firefox will run just fine on 64-bit Windows as a 32 bit process. Does that apply to Windows XP x64 because Windows XP's WOW64 isn't too powerful.
Is Firefox 4 compatible with Windows Server 2008?
OK, I think there is enough info in this bug now to fit the purpose outlined I'm Comment # > This doesn't block RC, > it blocks ship. The RC is meant to be the same as final. Blocking final blocks the RC. This bug should not be blocking20:final.
(In reply to comment #56) This is not a Firefox bug; it's a Mozilla.com bug. It doesn't add anything to Firefox itself so it doesn't affect the RC or final software release, just the web material to go with it. The only reason it's in Firefox:General with a blocking flag is to keep track of it because the website components can't block. I guess the proper thing to do would to be to add a flag for website components to block final shipping, but it doesn't come up all too often and those who are working on this and the RC know what's what. ;)
> those who are working on this and the RC know what's what. ;) Well they obviously do, since build1 of Firefox 4.0 RC1 is now on the server.  ftp://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/4.0rc1-candidates/  https://wiki.mozilla.org/Releases/Firefox_4.0rc1
And it's done on trunk. Need to push to staging, QA, and then live: http://viewvc.svn.mozilla.org/vc?view=revision&revision=84145 through http://viewvc.svn.mozilla.org/vc?view=revision&revision=84151
If any webdever wants to take this over for the stage+QA+production part, by all means take this from me. I don't have time this weekend to commit to getting this live so it will take until Monday if you wait for me (unless I have a wicked hangover on Sunday and push it from bed).
Also rev 84287 for an issue with poor interaction between blanket 4.0 redirects
There is still no status for Windows Server 2008.
The Linux section doesn't have hardware requirements. Is SSE2 required there too, for example?
Pushed live in http://viewvc.svn.mozilla.org/vc?view=revision&revision=84297. I think we can handle specific questions like comment 55 and 64 in a follow up.
For some reason, the Linux libstdc++ requirement (needs 4.3.0 or later) seems to have dropped off the "fixed" Firefox 4 requirements page. Any reason why this would be? Without it, CentOS 5.5 satisfies the current "fixed" requirements and yet it can't run Firefox 4 out of the box.
Reopening to at least get a response to comment 67. The libstdc++ requirement was in the agreed list of system requirements (as attached) but seems to have been dropped unintentionally when making the patch. But as this is live already, maybe that now has to be a follow up bug?
I don't know if this is to be mentioned here or not, the link at http://www.mozilla.com/en-US/firefox/4.0/system-requirements/index.html for firefox 4 download is pointing to http://www.mozilla.com/products/download.html?product=firefox-3.6.15&os=osx&lang=en-US
I guess it's normal since it's not released yet...
Pushed live the libstdc++ requirement in 84699. Please file a followup for further clarifications (and set this bug as blocking).
(In reply to comment #69) > I don't know if this is to be mentioned here or not, the link at > http://www.mozilla.com/en-US/firefox/4.0/system-requirements/index.html for > firefox 4 download is pointing to > http://www.mozilla.com/products/download.html?product=firefox-3.6.15&os=osx&lang=en-US Seems bad. Firefox home page http://www.mozilla.com/en-US/firefox/ Links to: http://www.mozilla.com/en-US/firefox/RC/ Links to: http://www.mozilla.com/en-US/firefox/all-beta.html Links to: http://www.mozilla.com/en-US/firefox/4.0rc1/releasenotes/ Links to: http://www.mozilla.com/en-US/firefox/system-requirements.html Big icon for "Firefox 4.0 Free Download" in upper right. Link goes to: Firefox 3.6.15 User scenario: Home page > What's this, Firefox 4 RC, wonder if it works on my system? >>>> OK, it does, let's download it > Click "Firefox 4.0" > Get Firefox 3.6.15.
It's now bug Bug 640461. Simpler scenario: Slashdot article > Release Notes > System Requirements > 3.6.15
Ah! Looks like the wrong revision might have got pushed live for the release notes, as trunk has the proper link.
Note comment 74 doesn't fix bug 640461, merely explains why http://www.mozilla.com/en-US/firefox/system-requirements.html showed up in the STR in comment 72
In reply to comment 65 and comment 71 > I think we can handle specific questions like comment 55 and 64 in a follow up. I filed bug 640783.
(In reply to comment #71) > Pushed live the libstdc++ requirement in 84699. Returning to verified fixed, with that.
As some people said, where's Windows Server 2008? (IE9 supports the system)
> As some people said, where's Windows Server 2008? (IE9 supports the system) See bug 640783.
OK, the scenario in Comment 72 is now fixed. Release notes links to http://www.mozilla.com/en-US/firefox/4.0rc1/system-requirements/ which has the 4.0RC download link and image. Also, http://www.mozilla.com/en-US/firefox/system-requirements.html now has the 4.0RC download link and image.