call for testing: firefox 4.0beta7
Hi,
after countless hours of hacking & building, i finally have a somewhat
working port of the last beta of firefox 4. It's running fine here on
i386 & i lightly tested it on amd64. Of course it doesn't build yet on
sparc64, didn't try ppc, upstream somewhat dropped official support
for anything !(i386|amd64), or at least it's not in the first
priorities. (anyway, openbsd is not an officially supported os..)
notes:
- most things seem to work fine, heavy js webapps feels better.
- Didn't try printing, it's probably still crashing.
- ipc/plugin sandboxing.. i don't know if it's useful, as apparently
it's useful only for binary plugins (flash/etc) which we don't have.
- port installs everything in lib/firefox-${VERSION}, that's what
upstream and other oses do. Stop the usr/local/mozilla-firefox
crazyness, and use the make install target upstream provides. I'd like
to do the same for other mozilla ports.
- i have to recheck if the xpm window icon hack is still needed
- test the build with systemwide cairo, should be okay as iirc they
bumped the internal version to 1.10.
- once systemwide sqlite is updated, try to use it with the #pragma hack
for secure_delete (and unlock-notify..)
- i didn't test java plugin compatibility, i doubt it'll work as it
already doesnt work with 3.6.
- xulrunner 2.0 needs to be polished/finished too
- it is intended to replace mozilla-firefox-3.6.x, so pkg_delete it
first. Backup your ~/.mozilla, or create a new profile for it.
- when it gets out of betas/rc, i'd like to avoid importing it into
www/firefox4 but rather update www/mozilla-firefox to this version.
we already have two versions of firefox in the portstree.. and it's a
mess to handle
- sparc64 & ppc support needs to be worked on
- various things i forgot.
Please test in various use cases. If it crashes where 3.6 didn't crash
before, build it with debug FLAVOR and try to report a useful trace
_upstream_ (as in http://bugzilla.mozilla.org, i wont handle everything)
get the port in my git repo (you'll need a -current portstree ofc, and
probably bumped ulimit -d if building as user) :
cd /usr/ports/mystuff/www/
git clone -b firefox-4 http://rhaalovely.net/git/mozilla-firefox/
cd mozilla-firefox && make install clean
or try those packages (built against somewhat -current, don't come crying
if it doesn't install on your box due to dependencies not matching) :
http://rhaalovely.net/stuff/i386/mozilla-firefox-4.0b7.tgz
http://rhaalovely.net/stuff/amd64/mozilla-firefox-4.0b7.tgz
Landry
Re: call for testing: firefox 4.0beta7
On Sat, 18 Dec 2010 18:14:36 +0100
Landry Breuil wrote:
> Hi,
Hi,
> after countless hours of hacking & building, i finally have a somewhat
> working port of the last beta of firefox 4. It's running fine here on
> i386 & i lightly tested it on amd64. Of course it doesn't build yet on
> sparc64, didn't try ppc, upstream somewhat dropped official support
> for anything !(i386|amd64), or at least it's not in the first
> priorities. (anyway, openbsd is not an officially supported os..)
>
> notes:
> - most things seem to work fine, heavy js webapps feels better.
> - Didn't try printing, it's probably still crashing.
> - ipc/plugin sandboxing.. i don't know if it's useful, as apparently
> it's useful only for binary plugins (flash/etc) which we don't have.
> - port installs everything in lib/firefox-${VERSION}, that's what
> upstream and other oses do. Stop the usr/local/mozilla-firefox
> crazyness, and use the make install target upstream provides. I'd like
> to do the same for other mozilla ports.
> - i have to recheck if the xpm window icon hack is still needed
> - test the build with systemwide cairo, should be okay as iirc they
> bumped the internal version to 1.10.
> - once systemwide sqlite is updated, try to use it with the #pragma hack
> for secure_delete (and unlock-notify..)
> - i didn't test java plugin compatibility, i doubt it'll work as it
> already doesnt work with 3.6.
> - xulrunner 2.0 needs to be polished/finished too
> - it is intended to replace mozilla-firefox-3.6.x, so pkg_delete it
> first. Backup your ~/.mozilla, or create a new profile for it.
> - when it gets out of betas/rc, i'd like to avoid importing it into
> www/firefox4 but rather update www/mozilla-firefox to this version.
> we already have two versions of firefox in the portstree.. and it's a
> mess to handle
> - sparc64 & ppc support needs to be worked on
> - various things i forgot.
>
> Please test in various use cases. If it crashes where 3.6 didn't crash
> before, build it with debug FLAVOR and try to report a useful trace
> _upstream_ (as in http://bugzilla.mozilla.org, i wont handle everything)
>
> get the port in my git repo (you'll need a -current portstree ofc, and
> probably bumped ulimit -d if building as user) :
>
> cd /usr/ports/mystuff/www/
> git clone -b firefox-4 http://rhaalovely.net/git/mozilla-firefox/
> cd mozilla-firefox && make install clean
Build and install OK.
I tested it with few sites, it's OK.
I'll use it in the next days instead of Midori to test ;).
> or try those packages (built against somewhat -current, don't come crying
> if it doesn't install on your box due to dependencies not matching) :
>
> http://rhaalovely.net/stuff/i386/mozilla-firefox-4.0b7.tgz
> http://rhaalovely.net/stuff/amd64/mozilla-firefox-4.0b7.tgz
>
> Landry
Thanks,
Remi.
Re: call for testing: firefox 4.0beta7
On Sat, Dec 18, 2010 at 06:14:36PM +0100, Landry Breuil wrote: > Hi, > > after countless hours of hacking & building, i finally have a somewhat > working port of the last beta of firefox 4. It's running fine here on > i386 & i lightly tested it on amd64. Of course it doesn't build yet on > sparc64, didn't try ppc, upstream somewhat dropped official support > for anything !(i386|amd64), or at least it's not in the first > priorities. (anyway, openbsd is not an officially supported os..) > > - sparc64 & ppc support needs to be worked on Good news, the port as-is built fine on my ibook and runs fine on the macmini.. so here's a package for mac lovers : http://rhaalovely.net/stuff/powerpc/mozilla-firefox-4.0b7.tgz If someone wants to try on other archs, i welcome patches :) Did some progress on sparc64 thx to mozilla bugzilla, but it doesn't build yet. Landry
Re: call for testing: firefox 4.0beta7
On 12/18/10 18:14, Landry Breuil wrote: > Hi, > > after countless hours of hacking & building, i finally have a somewhat > working port of the last beta of firefox 4. It's running fine here on > i386 & i lightly tested it on amd64. Of course it doesn't build yet on > sparc64, didn't try ppc, upstream somewhat dropped official support > for anything !(i386|amd64), or at least it's not in the first > priorities. (anyway, openbsd is not an officially supported os..) > > notes: > - most things seem to work fine, heavy js webapps feels better. > - Didn't try printing, it's probably still crashing. Printing support finally works @amd64, didn't try with gtk+-cups package installed, it seems to work very well. Thanks & Cheers Giovanni
Re: call for testing: firefox 4.0beta7
On Wed, Dec 22, 2010 at 03:44:47PM +0100, Giovanni Bechis wrote: > On 12/18/10 18:14, Landry Breuil wrote: > > Hi, > > > > after countless hours of hacking & building, i finally have a somewhat > > working port of the last beta of firefox 4. It's running fine here on > > i386 & i lightly tested it on amd64. Of course it doesn't build yet on > > sparc64, didn't try ppc, upstream somewhat dropped official support > > for anything !(i386|amd64), or at least it's not in the first > > priorities. (anyway, openbsd is not an officially supported os..) > > > > notes: > > - most things seem to work fine, heavy js webapps feels better. > > - Didn't try printing, it's probably still crashing. > Printing support finally works @amd64, didn't try with gtk+-cups package > installed, it seems to work very well. I think the crash happened only through gtk+2-cups, but as i barely even print i don't really remember. Landry
Re: call for testing: firefox 4.0beta7
On Wed, Dec 22, 2010 at 4:09 PM, Landry Breuil wrote: > On Wed, Dec 22, 2010 at 03:44:47PM +0100, Giovanni Bechis wrote: >> On 12/18/10 18:14, Landry Breuil wrote: >> > Hi, >> > >> > after countless hours of hacking & building, i finally have a somewhat >> > working port of the last beta of firefox 4. It's running fine here on >> > i386 & i lightly tested it on amd64. Of course it doesn't build yet on >> > sparc64, didn't try ppc, upstream somewhat dropped official support >> > for anything !(i386|amd64), or at least it's not in the first >> > priorities. (anyway, openbsd is not an officially supported os..) >> > >> > notes: >> > - most things seem to work fine, heavy js webapps feels better. >> > - Didn't try printing, it's probably still crashing. >> Printing support finally works @amd64, didn't try with gtk+-cups package >> installed, it seems to work very well. > > I think the crash happened only through gtk+2-cups, but as i barely even > print i don't really remember. My semi-uneducated guess is that cups printing is still broken. ciao, david
Re: call for testing: firefox 4.0beta7
On 12/22/10 16:09, Landry Breuil wrote: > On Wed, Dec 22, 2010 at 03:44:47PM +0100, Giovanni Bechis wrote: >> On 12/18/10 18:14, Landry Breuil wrote: >>> Hi, >>> >>> after countless hours of hacking & building, i finally have a somewhat >>> working port of the last beta of firefox 4. It's running fine here on >>> i386 & i lightly tested it on amd64. Of course it doesn't build yet on >>> sparc64, didn't try ppc, upstream somewhat dropped official support >>> for anything !(i386|amd64), or at least it's not in the first >>> priorities. (anyway, openbsd is not an officially supported os..) >>> >>> notes: >>> - most things seem to work fine, heavy js webapps feels better. >>> - Didn't try printing, it's probably still crashing. >> Printing support finally works @amd64, didn't try with gtk+-cups package >> installed, it seems to work very well. > > I think the crash happened only through gtk+2-cups, but as i barely even > print i don't really remember. > With Firefox 3.6 the crash happened without gtk+2-cups too @amd64, @i386 it has been "fixed" with latest libssl update (but I cannot test it atm). Cheers Giovanni
Re: call for testing: firefox 4.0beta7
On Wed, Dec 22, 2010 at 7:19 AM, Giovanni Bechis wrote: > On 12/22/10 16:09, Landry Breuil wrote: >> On Wed, Dec 22, 2010 at 03:44:47PM +0100, Giovanni Bechis wrote: >>> On 12/18/10 18:14, Landry Breuil wrote: Hi, after countless hours of hacking & building, i finally have a somewhat working port of the last beta of firefox 4. It's running fine here on i386 & i lightly tested it on amd64. Of course it doesn't build yet on sparc64, didn't try ppc, upstream somewhat dropped official support for anything !(i386|amd64), or at least it's not in the first priorities. (anyway, openbsd is not an officially supported os..) notes: - most things seem to work fine, heavy js webapps feels better. - Didn't try printing, it's probably still crashing. >>> Printing support finally works @amd64, didn't try with gtk+-cups package >>> installed, it seems to work very well. >> >> I think the crash happened only through gtk+2-cups, but as i barely even >> print i don't really remember. >> > With Firefox 3.6 the crash happened without gtk+2-cups too @amd64, @i386 > it has been "fixed" with latest libssl update (but I cannot test it atm). I have never had a problem printing through firefox3.6 (amd64/macppc). I don't use cups. read: it was never broken for me. --patrick
Re: call for testing: firefox 4.0beta7
> My semi-uneducated guess is that cups printing is still broken. Speaking only for the firefox mentioned in your subject line, AND with the current libssl, I'm happy to report that you are wrong: CUPS printing does work correctly in FF4.0 (at least on amd64). I have a few pieces of paper with marks on them to prove it.
Re: call for testing: firefox 4.0beta7
On Wed, Dec 22, 2010 at 11:46 PM, Ian Darwin wrote: > >> My semi-uneducated guess is that cups printing is still broken. > > Speaking only for the firefox mentioned in your subject line, AND with > the current libssl, I'm happy to report that you are wrong: CUPS > printing does work correctly in FF4.0 (at least on amd64). > I have a few pieces of paper with marks on them to prove it. gee! so good to know it. cheers, david
Firefox crashes when printing with CUPS, possible fix [Was Re: call for testing: firefox 4.0beta7]
On Wed, Dec 22, 2010 at 04:19:46PM +0100, Giovanni Bechis wrote: > On 12/22/10 16:09, Landry Breuil wrote: > > On Wed, Dec 22, 2010 at 03:44:47PM +0100, Giovanni Bechis wrote: > >> On 12/18/10 18:14, Landry Breuil wrote: > >>> Hi, > >>> > >>> after countless hours of hacking & building, i finally have a somewhat > >>> working port of the last beta of firefox 4. It's running fine here on > >>> i386 & i lightly tested it on amd64. Of course it doesn't build yet on > >>> sparc64, didn't try ppc, upstream somewhat dropped official support > >>> for anything !(i386|amd64), or at least it's not in the first > >>> priorities. (anyway, openbsd is not an officially supported os..) > >>> > >>> notes: > >>> - most things seem to work fine, heavy js webapps feels better. > >>> - Didn't try printing, it's probably still crashing. > >> Printing support finally works @amd64, didn't try with gtk+-cups package > >> installed, it seems to work very well. > > > > I think the crash happened only through gtk+2-cups, but as i barely even > > print i don't really remember. > > > With Firefox 3.6 the crash happened without gtk+2-cups too @amd64, @i386 > it has been "fixed" with latest libssl update (but I cannot test it atm). Are you sure it's been "fixed" by the openssl update ? There's been progress upstream, so if anyone still sees the crashes with gtk+2-cups and ffx 3.6 or ffx4, please try the patch from https://bugzilla.mozilla.org/show_bug.cgi?id=573039 https://bug573039.bugzilla.mozilla.org/attachment.cgi?id=499233 (as pointed out by aja@) Landry
