You are not logged in.
Pages: 1
Hi,
I want to recompile firefox to get rid of the Gran Paradiso bull.
I went to this page, but it appears to be for a much earlier version of firefox:
http://wiki.archlinux.org/index.php/Firefox
for example in the mozconfig:
in the current build package we have:
ac_add_options --with-branding=browser/branding/unofficial
and in the instructions it says to add this:
ac_add_options --enable-official-branding
Which is different than what's in the PKGBUILD.
Also it says to replace lines like this:
convert ${startdir}/src/mozilla/browser/app/default.xpm ${startdir}/pkg/usr/share/pixmaps/firefox.png
with this:
convert ${startdir}/src/mozilla/dist/branding/default.xpm ${startdir}/pkg/usr/share/pixmaps/firefox.png
but there are no convert lines.
Has anyone recompiled to get the branding issue resolved for the latest version of FF?
Offline
Hey,
it was annoying for me as well. But there is no need to recompile it, just use thunderbrand script from archlinuxfr repo.
Easy as that.
Bye xenol
Offline
Well ... you can always get the one from mozilla's page (pre-compiled), write a PKGBUILD and install it into your system.
Should be simple enough after you find out where to put all the stuff (took me a long while to figure out all the stuff I needed to do the first time).
Latest firefox is here (as of the time of writing) http://releases.mozilla.org/pub/mozilla … .6.tar.bz2 if you want to make the pkgbuild.
Last edited by R00KIE (2009-02-18 00:29:40)
R00KIE
Tm90aGluZyB0byBzZWUgaGVyZSwgbW92ZSBhbG9uZy4K
Offline
Has anyone recompiled to get the branding issue resolved for the latest version of FF?
Unless I'm misunderstanding something, that's what firefox-branded does. I don't know if it compiles properly since I don't use it, but I'm pretty sure it does exactly what you're trying to do.
Last edited by tdy (2009-02-18 01:30:50)
Offline
Using firebrand to rebrand Firefox is far faster than compiling firefox-branded.
Offline
Using firebrand to rebrand Firefox is far faster than compiling firefox-branded.
Depends on how fast your system is. I just upgraded to Phenom II 3.0 gzh and firefox compiles in a couple of minutes :-)
Offline
have you tried firefox-pgo from aur? takes a while to compile but it's branded and (from the description):
Mozilla Firefox, the way it should be packaged
(https://developer.mozilla.org/en/Buildi … timization).
Replacement for firefox-spookyet. XULRunner independent
"I know what you're thinking, 'cause right now I'm thinking the same thing. Actually, I've been thinking it ever since I got here:
Why oh why didn't I take the BLUE pill?"
Offline
have you tried firefox-pgo from aur? takes a while to compile but it's branded and (from the description):
Mozilla Firefox, the way it should be packaged
(https://developer.mozilla.org/en/Buildi … timization).
Replacement for firefox-spookyet. XULRunner independent
Compiling it now, taking a lot longer than the standard package, but from what I have been reading on the PGO builds they are supposed to
be super fast.
Thanks for the tip :-)
Offline
Compiling it now, taking a lot longer than the standard package, but from what I have been reading on the PGO builds they are supposed to
be super fast.Thanks for the tip :-)
i've been using the spookyet package before this one, this one took twice as long to compile (i believe it actually does compile ff twice).. after the long compile, it does seem snappier to me, and havent had any issues whatsoever. plus i like watching the compiles, i find them very soothing...
that's just me though
it's like watching the kernel compile, you're in the matrix now..
Last edited by toxygen (2009-02-18 02:50:06)
"I know what you're thinking, 'cause right now I'm thinking the same thing. Actually, I've been thinking it ever since I got here:
Why oh why didn't I take the BLUE pill?"
Offline
have you tried firefox-pgo from aur? takes a while to compile but it's branded and (from the description):
Mozilla Firefox, the way it should be packaged
(https://developer.mozilla.org/en/Buildi … timization).
Replacement for firefox-spookyet. XULRunner independent
That's PGO... it takes me over an hour for my profile-guidied optimization of FF 3.1b2 (it's worth it).
But a regualr build of firefox takes about 15 - 20 minutes for me.... if that.... more like 10 - 15 minutes.
You could of compiled the AUR firefox-branded package in the time that it took to participate in this thread.
Firefox-branded: http://aur.archlinux.org/packages.php?ID=18019
Last edited by methuselah (2009-02-18 03:07:08)
Offline
toxygen wrote:have you tried firefox-pgo from aur? takes a while to compile but it's branded and (from the description):
Mozilla Firefox, the way it should be packaged
(https://developer.mozilla.org/en/Buildi … timization).
Replacement for firefox-spookyet. XULRunner independentCompiling it now, taking a lot longer than the standard package, but from what I have been reading on the PGO builds they are supposed to
be super fast.Thanks for the tip :-)
Hey snorkel..... you should try the 3.1b2-PGO ..... it's really fast and has been stable for me. Just add the about:config setting of "extensions.checkCompatibility = false". Also makesure your C[XX]FLAGS and MAKEFLAGS are set in the /etc/makepkg.conf file.
Flash64 and java64 have worked perfectly for me as well as my mozplugger plugin.
Last edited by methuselah (2009-02-18 03:15:23)
Offline
Ranguvar wrote:Using firebrand to rebrand Firefox is far faster than compiling firefox-branded.
Depends on how fast your system is. I just upgraded to Phenom II 3.0 gzh and firefox compiles in a couple of minutes :-)
Firebrand goes in seconds.
Anyways, about PGO, IMHO it's a placebo (Are you using the 3.1 beta? It has a lot of speed improvements by itself). I'll benchmark some time though. I just don't see how compile-time optimizations can help an app like Firefox any significant amounnt in terms of speed... compiling with -Os actually helps in terms of RAM usage, though.
EDIT: Maybe not a placebo... http://cybernetnews.com/2008/02/25/fire … s-a-boost/ I'd still prefer real-life page load benchmarks though. And LOL @ IE.
Last edited by Ranguvar (2009-02-18 04:05:54)
Offline
snorkel wrote:Ranguvar wrote:Using firebrand to rebrand Firefox is far faster than compiling firefox-branded.
Depends on how fast your system is. I just upgraded to Phenom II 3.0 gzh and firefox compiles in a couple of minutes :-)
Firebrand goes in seconds.
Anyways, about PGO, IMHO it's a placebo (Are you using the 3.1 beta? It has a lot of speed improvements by itself). I'll benchmark some time though. I just don't see how compile-time optimizations can help an app like Firefox any significant amounnt in terms of speed... compiling with -Os actually helps in terms of RAM usage, though.
EDIT: Maybe not a placebo... http://cybernetnews.com/2008/02/25/fire … s-a-boost/ I'd still prefer real-life page load benchmarks though. And LOL @ IE.
I've never benchmarked anything so I'm not sure about your placebo theory..... but I do use a completely source compiled Archlinux. All my apps are compiled with the best C[XX]FLAGS for my processor using pacbuilder-svn, and I make my own compiled AMD64 kernel26..... so I also build Firefox with my C[XX]FLAGS..... and why not use the PGO build to optimize the best I possibly can.
I know that the regular 3.1b2 and 3.2alpha's are very fast (I use them on Vista), so I would be interested in seeing some benchmarks also. Since I'm doing a "gentoo-ish" Arch system using only ABS and AUR then I would be building the FF 3.1b2 anyway so why not just go for PGO also.
Offline
You recompile everything, but didn't benchmark? x.x (Not an insult, just saddened that we still don't have hard data on the usefulness of that, and IMO it's a little overboard for not benchmarking at all)
Kinda offtopic, but IMO if you're going to recompile, the best optimizations are _not_ in C(XX)FLAGS, but in LDFLAGS. Any C(XX)FLAGS beyond "-march=native -O2 -pipe" just caused trouble for me, but LDFLAGS can cause apps to start much faster. Ex: "-Wl,-O1 -Wl,--as-needed -Wl,--hash-style=gnu -Wl,--sort-common" works without much trouble, as long as you recompile everything with that.
Offline
Hey snorkel..... you should try the 3.1b2-PGO ..... it's really fast and has been stable for me. Just add the about:config setting of "extensions.checkCompatibility = false". Also makesure your C[XX]FLAGS and MAKEFLAGS are set in the /etc/makepkg.conf file.
Flash64 and java64 have worked perfectly for me as well as my mozplugger plugin.
I hadnt heard about that checkCompatibility option.. are there any "known" extensions that break? last time i tried a 3.1 browser half my needed extensions (tabmix plus, foxmarks, downloadthemall) werent working too well. I'll have to check it out when i get home
"I know what you're thinking, 'cause right now I'm thinking the same thing. Actually, I've been thinking it ever since I got here:
Why oh why didn't I take the BLUE pill?"
Offline
You recompile everything, but didn't benchmark? x.x (Not an insult, just saddened that we still don't have hard data on the usefulness of that, and IMO it's a little overboard for not benchmarking at all)
Kinda offtopic, but IMO if you're going to recompile, the best optimizations are _not_ in C(XX)FLAGS, but in LDFLAGS. Any C(XX)FLAGS beyond "-march=native -O2 -pipe" just caused trouble for me, but LDFLAGS can cause apps to start much faster. Ex: "-Wl,-O1 -Wl,--as-needed -Wl,--hash-style=gnu -Wl,--sort-common" works without much trouble, as long as you recompile everything with that.
I use : "-march=athlon64 -msse3 -Os -pipe" and it works good for me. I don't feel I need to benchmark. I figure that the preformance gain is there, and even if it's not much, it's still something that I wanted to try..... like a Gentoo install except using Arch ABS and AUR.
Offline
methuselah wrote:Hey snorkel..... you should try the 3.1b2-PGO ..... it's really fast and has been stable for me. Just add the about:config setting of "extensions.checkCompatibility = false". Also makesure your C[XX]FLAGS and MAKEFLAGS are set in the /etc/makepkg.conf file.
Flash64 and java64 have worked perfectly for me as well as my mozplugger plugin.
I hadnt heard about that checkCompatibility option.. are there any "known" extensions that break? last time i tried a 3.1 browser half my needed extensions (tabmix plus, foxmarks, downloadthemall) werent working too well. I'll have to check it out when i get home
3.1 does truly break a lot of extensions, as that framework has been revamped quite a bit IIRC. But checkCompatibility does make some others work.
Offline
There is firefox-pgo-beta on AUR. It compiles 3.1b2 with CFLAGS set by user and PGO
Proud ex-maintainer of firefox-pgo
Offline
There is firefox-pgo-beta on AUR. It compiles 3.1b2 with CFLAGS set by user and PGO
Your Firefox-PGO 3.1b2 has worked perfectly for me. Thank you for maintaining that package blasse.
snorkel wrote:Has anyone recompiled to get the branding issue resolved for the latest version of FF?
Unless I'm misunderstanding something, that's what firefox-branded does. I don't know if it compiles properly since I don't use it, but I'm pretty sure it does exactly what you're trying to do.
This would also work if you didn't want to build the PGO..... and if you needed to build a different package than the PGO or the firefox-branded..... for example this package: http://aur.archlinux.org/packages.php?ID=18090
...... then you can just modify the PKGBUILD (or any other unbranded Firefox PKGBUILDs) to replace this line:
install -m644 ${srcdir}/mozilla/browser/branding/unofficial/default48.png ${pkgdir}/usr/share/pixmaps/firefox.png || return 1
with this line:
install -m644 ${srcdir}/mozilla/other-licenses/branding/firefox/default48.png ${pkgdir}/usr/share/pixmaps/firefox.png || return 1
and also edit the mozconfig file to change this line:
ac_add_options --with-branding=browser/branding/unofficial
to this:
ac_add_options --enable-official-branding
That has worked for me.
Last edited by methuselah (2009-03-01 18:56:52)
Offline
As Ranguvar said: Save yourselves some time. Use Firebrand and Thunderbrand!! They are quick and painless! Here's the link: http://bbs.archlinux.org/viewtopic.php?id=44320
Offline
Pages: 1