Does the Gentoo build process start afresh on each compile ? The
conventional wisdom is to remove all the old compilation products when
you change the compiler. For a manual build I'd just delete the object
directory, or use "make -f client.mk clean".
C F wrote:
> Does the Gentoo build process start afresh on each compile ?
Gentoo's own port of firefox doesn't suffer from this problem, I'm
happy to say. I have the official gentoo port of firefox installed
(in a different directory, of course) which works perfectly.
I'm referring only to my own personal build of firefox-trunk which I
update from CVS every morning. I don't want anyone to think this is a
gentoo problem (unless it's a bug in their gcc-3.4.4 port, which is far
beyond my ability to judge.) That's really my motive for posting this
question -- if there *is* a bug in gcc-3.4.4 I'd like to report it to
someone who can fix it.
> conventional wisdom is to remove all the old compilation products when
> you change the compiler. For a manual build I'd just delete the object
> directory, or use "make -f client.mk clean".
Yes, I learned that the hard way after filing several bogus bugreports
on bugzilla. The first thing I do when I encounter a bug is to delete
the objdirectory and start over -- this fixes the majority of problems
that bite me on my daily journey into bugland ;o)
Uli Link wrote:
> walt schrieb:
>> I just tried building SeaMonkey-trunk with 3.4.4 and all works
>> okay, so this seems to be specific for FF, and has something to
>> do with java applets AFAICT.
> There are binary incompatibilities with different versions of GCC an C++
> shared libs. So if you use Plugin's you should use the same version as
> your plugins are compiled with.
Alas, I gave you false information. Here is that famous phrase:
"On further investigation..." I determined that my problem does not
depend on any plugin. I deleted all plugins and still had the same
problem with that one website. I fear that I will need to learn a lot
more about gdb than I ever wanted to know :o(
I'm also having trouble with my own Firefox 1.5 builds and gcc-3.4.4.
(using Gentoo as well) - at least I think it has something to do with
gcc because my previous gcc-3.3.6 build works fine.
Please have a look at this thread here for further details:
> I build firefox from CVS every day on my gentoo-stable
> machine. The gentoo folks just marked gcc-3.4.4 as stable,
> so I tried it. I had no problems with it except for firefox.
> Firefox compiles with no errors, but when I run the program
> it freezes during startup and consumes 99% of cpu until I kill
> the firefox-bin process.
> Has anyone successfully built firefox with gcc-3.4.4?
> Hi walt,
> I'm also having trouble with my own Firefox 1.5 builds and gcc-3.4.4.
> (using Gentoo as well)...
I had disk space to spare, so I installed another instance of gentoo
and made it an 'unstable' test box by adding the ~x86 useflag.
The gcc on the 'unstable' box is 3.4.5, which builds firefox with
no problems, and firefox runs normally.
I have considered emerging gcc-3.4.5 on my stable box as a test --
but the problems with conflicting versions of libstdc++ have put me
off the idea for now. If you would like to suffer the headaches,
however, I'd be interested in your results ;o)
When I built FF on my new 'unstable' test box I was careful to
do all the 'right stuff' like revdep-rebuild and use gcc-config
to set the profile to vanilla-gcc, rebuild libtool, and run
I'm not willing to do all that invasive stuff on my 'stable' box
until I'm confident that it won't be a catastrophe. The test box,
OTOH, who cares! As it happens, the 'unstable' box is working at
least as well as the 'stable' box. But that could change tomorrow.
If you want to use your new compiler, you really *must* recompile
every package which is linked against libstdc++. That's why I
ran revdev-rebuild against the new libstdc++. It's really quite
a PIA and once you do it, it's an equal PIA to go back to the
mozilla-builds mailing list
[hidden email] http://mail.mozilla.org/listinfo/mozilla-builds