| Summary: | www-client/mozilla-firefox-1.5.0.7 doesn't compile | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Erik Logtenberg <erik> |
| Component: | Current packages | Assignee: | Gentoo's Team for Core System packages <base-system> |
| Status: | RESOLVED WORKSFORME | ||
| Severity: | normal | ||
| Priority: | High | ||
| Version: | 2006.1 | ||
| Hardware: | x86 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: |
Output of emerge --info
Strace (-f) log of emerge mozilla-firefox # strace -f emerge --oneshot dev-libs/nss > strace.log 2>&1 |
||
|
Description
Erik Logtenberg
2006-09-19 10:07:33 UTC
Created attachment 97436 [details]
Output of emerge --info
This is the output of emerge --info on my system.
firefox-1.5.0.7 certainly builds with gcc-4.1.1; something else is wrong on your system That is certainly true. Any ideas on how I can gather more information on this issue so it can be tracked down? I do believe the upgrade to gcc-4.1.1 has something to do with it, since firefox (and thunderbird and enigmail) have stopped compiling during the "emerge -e world" that was part of the migrationprocess. However this seems to happen only in certain circumstances. Does my "emerge --info" output provide enough information on these circumstances for anyone to be able to reproduce it? post the actual build errors How do I get to see 'the actual build errors'? I pasted the output of emerge literally, except where the (...) are. You see that the "!!! ERROR: www-client/mozilla-firefox-1.5.0.7 failed." line follows directly after the "./configure --prefix=/usr --host=i686-pc-linux-gnu --mandir=/usr/share/man --infodir=/usr/share/info --datadir=/usr/share --sysconfdir=/etc --localstatedir=/var/lib --build=i686-pc-linux-gnu" part. There is no configure output that I left out, it simply returns the error before it actually seems to begin configuring. *** Bug 148215 has been marked as a duplicate of this bug. *** Your system appears to have completely messed up autotools or whatnot, this has nothing to do w/ gcc version nor w/ firefox (same with Bug 148215) That may be true, but I can still compile any other package that I want. Even if gcc-4.1.1 is not the cause of this bug, it's still a fact that I was able to compile firefox before I upgraded to gcc-4.1.1, and during upgrade to gcc-4.1.1. it stopped working. I followed the steps in the howto: 1) Use gcc-config to select gcc-4.1.1 (and do source /etc/profile) 2) emerge libtool 3) emerge -e system 4) emerge -e world With step 4, firefox, thunderbird and libstdc++-v3 didn't compile. I used 'emerge --resume --skipfirst' for each of them, which continued fine. 5) build a new kernel 6) reboot After reboot, I tried again if I could emerge firefox and/or thunderbird, which was still not the case. At this moment I'm still using the firefox that has been compiled with gcc-3.4.6 and this still works like a charm. However I would like to upgrade firefox to the latest version, since there are some security issues with the version I still use. Therefore I'd like to be able to compile it again. Do you have any thoughts on how I can assert which package is messed up, and how to fix this? And how it is possible that I can still compile any other package? > Your system appears to have completely messed up autotools or whatnot
Could you try to specify the "whatnot" please? I checked autotoolset, and it's not even installed on my system. I tried installing dev-util/autotoolset-0.11.4-r1 (and installed app-arch/sharutils-4.2.1-r11 as a dependancy). Although emerge'ing autotoolset works fine, it didn't fix the firefox issue.
Created attachment 97567 [details]
Strace (-f) log of emerge mozilla-firefox
This is a snippet from the strace log. I think this is where it goes wrong. It shows the "/configure ..."-line and the "!!! ERROR..."-line.
You can see stuff like:
[pid 20092] <... getpeername resumed> 0xbf937044, [128]) = -1 ENOTCONN (Transport endpoint is not connected)
[pid 19971] <... rt_sigprocmask resumed> [], 8) = 0
[pid 20092] write(3, "\25\3\1\0 /\10}X\246\27\343e\202\2656T\371*\261\235\370"..., 37 <unfinished ...>
in there. There's already a lot of this above the "./configure"-line, so maybe the fault is has already occurred earlier on.
I'd like your opinion on this.
go into the portage dir and run ./configure yourself with the same options and see what happens Okay, I tried running that ./configure statement manually, and guess what.. it works like a charm. Moreover, I've found out that when I run the emerge process step by step, as follows: # ebuild /usr/portage/www-client/mozilla-firefox/mozilla-firefox-1.5.0.7.ebuild unpack # ebuild /usr/portage/www-client/mozilla-firefox/mozilla-firefox-1.5.0.7.ebuild compile # ebuild /usr/portage/www-client/mozilla-firefox/mozilla-firefox-1.5.0.7.ebuild install # ebuild /usr/portage/www-client/mozilla-firefox/mozilla-firefox-1.5.0.7.ebuild qmerge Then it compiles and merges just fine! However when I attempt to use 'emerge' like so: # emerge mozilla-firefox it fails, as described in the posts above. I notice the same issue with libstdc++-v3. pust the full strace; you snipped too much Created attachment 99140 [details]
# strace -f emerge --oneshot dev-libs/nss > strace.log 2>&1
Hi, The problem seems to be more a portage-issue than a Firefox issue. The same thing happens for instance to dev-libs/nss. When I run a "emerge -uD --newuse world", most packages compile fine, but some do not. dev-libs/nss is one of them. To show this, I've attached the output of: # strace -f emerge --oneshot dev-libs/nss > strace.log 2>&1 Please note that I'm perfectly able to get dev-libs/nss installed by using the 'ebuild' tool, and stepping through compile, install and qmerge manually: # ebuild /usr/portage/dev-libs/nss/nss-3.11.3-r1.ebuild compile # ebuild /usr/portage/dev-libs/nss/nss-3.11.3-r1.ebuild install # ebuild /usr/portage/dev-libs/nss/nss-3.11.3-r1.ebuild qmerge And voila, it's installed just fine. Exactly the same thing with mozilla-firefox and a couple of other packages, and perfectly reproducable. It only affects a small portion of packages, most don't have this problem. (Perhaps "Applications" isn't quite the right Component for this bug?) Kind regards, Erik Logtenberg. (In reply to comment #15) > Hi, > > The problem seems to be more a portage-issue than a Firefox issue. The same > thing happens for instance to dev-libs/nss. When I run a "emerge -uD --newuse > world", most packages compile fine, but some do not. dev-libs/nss is one of > them. To show this, I've attached the output of: > > # strace -f emerge --oneshot dev-libs/nss > strace.log 2>&1 > > Please note that I'm perfectly able to get dev-libs/nss installed by using the > 'ebuild' tool, and stepping through compile, install and qmerge manually: > > # ebuild /usr/portage/dev-libs/nss/nss-3.11.3-r1.ebuild compile > # ebuild /usr/portage/dev-libs/nss/nss-3.11.3-r1.ebuild install > # ebuild /usr/portage/dev-libs/nss/nss-3.11.3-r1.ebuild qmerge > > And voila, it's installed just fine. > Exactly the same thing with mozilla-firefox and a couple of other packages, and > perfectly reproducable. It only affects a small portion of packages, most don't > have this problem. > (Perhaps "Applications" isn't quite the right Component for this bug?) > > Kind regards, > > Erik Logtenberg. > I am not sure what you expect gentoo to do here. It works fine for all of us. I suggest closing bug wfm, unless a dev can actually dupilicate this bug. Resolving WORKSFORME as per comment #16 - please provide additional information to reproduce this. |
