Opened 11 years ago

Closed 11 years ago

Last modified 11 years ago

#3852 closed enhancement (fixed)

firefox-21.0

Reported by: ken@… Owned by: Armin K
Priority: normal Milestone:
Component: BOOK Version: SVN
Severity: normal Keywords:
Cc:

Description

new version, and downloadable from SF (yeugh!) http://sourceforge.net/projects/portableapps/files/Source/Firefox/

The page at releases.mozilla.org now says:

Mozilla Project Mirror

As the Mozilla Foundation has decided to move towards using Content Distribution Networks (CDNs) to distribute Mozilla binaries, the recommended source for Firefox is http://getfirefox.com (all Mozilla products are available at http://www.mozilla.org/products). We are freeing the space on the Open Source Mirror in order to make room for other Open Source projects in need of mirroring services.

Attachments (1)

firefox-packager-build.patch (1.0 KB ) - added by ken@… 11 years ago.

Download all attachments as: .zip

Change History (11)

comment:1 by ken@…, 11 years ago

Current versions of deps seem adequate [ I'm posting this from a DESTDIR install of firefox-without-xulrunner ] but I've so far got two issues:

  1. in both firefox and xulrunner, WITHOUT setting MOZ_MAKE_FLAGS it appears to be using all 4 cores of my Phenom [ 13.5 SBU for the whole firefox-only build/install ].
  2. for me, xulrunner fails to build -

/scratch/ken/mozilla-release/js/src/jsclone.cpp:293:5: note: in expansion of macro ‘JS_STATIC_ASSERT’

JS_STATIC_ASSERT(sizeof(uint64_t) % sizeof(T) == 0);

In file included from /scratch/ken/mozilla-release/js/src/jsapi.h:35:0,

from /scratch/ken/mozilla-release/js/src/jsclone.h:9, from /scratch/ken/mozilla-release/js/src/jsclone.cpp:8:

./../../dist/include/js/Vector.h: In instantiation of ‘T* js::Vector<T, MinInlineCapacity, AllocPolicy>::beginNoCheck() const [with T = js::Shape*; long unsigned int N = 8ul; AllocPolicy = js::TempAllocPolicy]’: ./../../dist/include/js/Vector.h:581:32: required from ‘js::Vector<T, MinInlineCapacity, AllocPolicy>::~Vector() [with T = js::Shape*; long unsigned int N = 8ul; AllocPolicy = js::TempAllocPolicy]’ /scratch/ken/mozilla-release/js/src/jsapi.h:314:7: required from here ./../../dist/include/js/Vector.h:279:8: internal compiler error: Segmentation fault

T *beginNoCheck() const {

DWARF underflow in .debug_info at 4 unrecognized DWARF version in .debug_info at 4 Please submit a full bug report, with preprocessed source if appropriate. See <http://gcc.gnu.org/bugs.html> for instructions.

Maybe this is a gcc-4.8 issue, not sure. I don't understand what is objected to, or required, here.

comment:2 by ken@…, 11 years ago

Hmm, on LFS-7.3 [ gcc-4.7.2 ] it's worse - xulrunner fails to build, but I've got two unkillable cc1plus jobs still taking about 100% cpu each.

No idea if it was the same error, but can someone remind me why we bother to build xulrunner now, please ? I know that it used to appear to be the right thing to do (looked as if e.g. thunderbird would one day be able to use it), and my measurements appeared to show it was quicker to build both (might have been mistaken), but apart from firefox, what uses it ? I've a vague memory of java perhaps needing it ?

comment:3 by Armin K, 11 years ago

Owner: changed from blfs-book@… to Armin K
Status: newassigned
 ./../../dist/include/js/Vector.h:279:8: internal compiler error:
 Segmentation fault

c++ compiler segfaulted. I've experienced same issue several times when system was under heavy load (using all of the cores). Restarting "make" fixed the issue.

I've built Xulrunner last night, but I couldn't get Firefox to build against it - I was hit by a bug that dates to 2004 "Failed to connect to .... service.".

Using gcc 4.8.1 snapshot -> gcc version 4.8.1 20130502 (prerelease)

comment:4 by ken@…, 11 years ago

Thanks for those comments - if I uncomment MOZ_MAKE_FLAGS in the mozconfig, and set that value to -j1 it does build [ 44 SBU for xulrunner ] : looks as if the default is now to use all cores ? Slightly irrelevant since you can't build firefox against it.

I've only recently resumed using -j4 on this machine - a year or so ago all sorts of things used to break randomly. With my last build on this box (LFS-svn-20130424) -j4 worked everywhere. At the moment I'm doing a j1 build of firefox to compare the time.

comment:5 by ken@…, 11 years ago

Just installed firefox without xulrunner after seeing the error. But maybe this patch will help (seems to have an extra directory in it!) ? http://pkgs.fedoraproject.org/cgit/firefox.git/plain/firefox-packager-build.patch

comment:6 by Armin K, 11 years ago

Mhm, It would help me if you added a patch a bit earlier. I just removed the whole Xulrunner build after another failure ... I'll try to build it again (for the 4th time).

by ken@…, 11 years ago

comment:7 by ken@…, 11 years ago

Sorry, I didn't find it until after I'd seen the error message and spent some time searching. In case you haven't started, I'll attach my version of it. The name doesn't meet BLFS standards, call it whatever you like. I've now used it to build ff-on-xul on this other box, but I think I'll go back to plain ff-only in future - it's quicker!

comment:8 by Armin K, 11 years ago

Woot, the patch worked! I've installed Firefox+Xulrunner now. I'll create a sed from the patch - no need to add a patch for one liner fix.

comment:9 by Armin K, 11 years ago

Resolution: fixed
Status: assignedclosed

Fixed in r11178

comment:10 by bdubbs@…, 11 years ago

Milestone: current

Milestone current deleted

Note: See TracTickets for help on using tickets.