Package Details: firefox-beta 70.0b14-1

Git Clone URL: https://aur.archlinux.org/firefox-beta.git (read-only)
Package Base: firefox-beta
Description: Standalone web browser from mozilla.org - Beta
Upstream URL: https://www.mozilla.org/en-US/firefox/channel/#beta
Keywords: browser gecko web
Licenses: GPL, MPL, LGPL
Conflicts: firefox-beta-bin
Provides: firefox=70.0b14
Submitter: argymeg
Maintainer: FredBezies
Last Packager: FredBezies
Votes: 13
Popularity: 0.000030
First Submitted: 2015-10-05 23:34
Last Updated: 2019-10-12 12:09

Dependencies (35)

Required by (189)

Sources (4)

Latest Comments

« First ‹ Previous ... 2 3 4 5 6 7 8 Next › Last »

AnAkkk commented on 2015-11-05 10:17

This should now depend on gtk3.

Det commented on 2015-10-28 15:44

The GUI is probably slightly faster due to the use of system-level libraries (my own experience as well).

Generally, the optimizations alone you get with things like -march=native only yield you ~5% performance boost:

- http://serverfault.com/questions/550350/performance-difference-between-compiled-and-binary-linux-distributions-packages
- http://askubuntu.com/questions/102391/is-it-better-to-compile-from-source-or-to-install-from-a-deb-package

If you seriously want to compile Firefox yourself (and don't own a fast CPU), even with ccache, I'd suggest only doing it for the stable releases. The Beta channel updates about every 3-4 days, and you'll probably see yourself returning to the -bin version later on anyway.

argymeg commented on 2015-10-28 12:18

The most tangible benefit is using a number of system libraries rather than those bundled with Firefox, which makes for a more up-to-date and streamlined system. Also, the official builds still use GStreamer 0.10, a long-deprecated version. Using GStreamer 1.0 fixes several (even if somewhat obscure) media playback issues for me.

As far as performance gains go, whether recompiling packages for modern systems is worth the time and effort is often debated and generally differs from case to case. I haven't done any benchmarks which would be the obvious way to decide on this, yet my own builds do feel faster for me, entirely subjectively.

dxxvi commented on 2015-10-28 05:23

Could somebody tell what the noticeable profits are from building firefox from source?

argymeg commented on 2015-10-11 23:23

My bad, it seems it was some issue with my setup, building in a clean chroot (https://wiki.archlinux.org/index.php/DeveloperWiki:Building_in_a_Clean_Chroot#Classic_Way) worked fine with the proposed patch. Updated.

argymeg commented on 2015-10-10 11:16

The patch doesn't solve the issue for me. I have reported it to mozilla, package updated and the version dependency stays for now.

Det commented on 2015-10-09 13:33

No, but he already posted a working patch at the end of the report. I missed it at first too.

argymeg commented on 2015-10-09 12:34

I have to admit I hadn't thought of it this way at all. Fixed and thank you. I will follow the bug and probably patch it manually if it isn't fixed by the next release.

Det commented on 2015-10-09 09:23

They don't, but that's the point. If you can't build with A, version B, then you need to disallow that dependency in the PKGBUILD. Even if you can't find "pure" 2.6 anywhere, there's no reason not to warn the user that you know for sure the build will fail (unless you downgrade manually). Why let the user build through then?

E: There's actually already a bug report and a patch for 2.6.1: https://bugzilla.mozilla.org/show_bug.cgi?id=1213186

argymeg commented on 2015-10-09 07:53

I already tried that but it fails with "Missing dependencies: freetype2<2.6.1", which actually seems quite reasonable given that the repos don't keep the old versions around, unless I'm missing something here.

You're probably right, I think changing it to "build from source" would be less ambiguous.