[blfs-dev] FireFox 29.0.1

Ken Moffat zarniwhoop at ntlworld.com
Mon Jun 2 16:19:16 PDT 2014


On Tue, Jun 03, 2014 at 10:29:55AM +1200, Christopher Gregory wrote:
> 
> Well I don't know if it will ever be narrowed down, as it appears to be
> one of these random things that affect some and not others.
> 
> I have a:
> 
> model name      : Genuine Intel(R) CPU           T2300  @ 1.66GHz
> centino duo 32 bit processor.
> 
> Maybe this weirdness is in some way unique to a 32bit system?
> 
 For the two occasions that I can remember firefox crashing at
sourceforge, I'm 99% certain that I was on x86_64.  But for me,
firefox has always crashed from time to time - and been ok when I
retry the page I was trying to read.

> As for using the system cairo, It states in the mozconfig file that if we
> want to use the system cairo to uncomment the line.  Also on the cairo
> installation page it states that in order to use cairo with firefox the
> enable-tee option must be used.
> 
 Personally, I think people ought to use the system versions of
libraries whenever possible.  When I _measure_ firefox and xulrunner
for BLFS, I use the options in the book.  But before that, I build
and run it/them with system libraries and check that things seem to be
working.

> If we are NOT meant to be using the system cairo with firefox, perhaps a
> note in the mozconfig woud be appropriate, along with the removal of the
> enable-tee option on the cairo page?
> 

 Armin has a different view, but he is no longer editing BLFS and at the
moment I do not share his view.  For myself, I have not seen anything
which indicates there is a common problem with firefox using system
cairo.  As always, much can change from one user to another and
problems might be down to specific sites.

 I would not object to adding a note in the mozconfigs along the
lines of "some people report that this can make firefox crash at
certain sites", but I'm uncertain about what wording I would be
willing to accept for that, and ideally it ought to fit into at most
two lines!

 It isn't a problem I can reliably re-create, so I'm reluctant to do
this.  One of the problems with many older sets of notes is that
they become out of date but get carried forward regardless.
Presumably, when the next stable version of cairo gets released, the
problem will disappear.

 I suppose we could commit the upstream cairo patch, but since I
can't recreate the problem I'm not in a rush to do that (too much
else - non-BLFS - to do).
> I only meantion that regarding cairo as I have just read through all the
> comments and noticed one person said to use the bunndled cairo.
> 
> The thing is, that so far, with that patch applied, firefox is at this
> stage behaving itself.
> 
> Sorry about resending, but I am using webmail and accidently clicked on
> send before finishing the email.
> 
> Regards,
> 
> Christopher.
> 

 What really puzzles me is why some firefox issues seem to affect
very few people.  On one of my older systems I had to change
_something_ in the last couple of months when I upgraded firefox (I
no longer remember the details, but I think it was a specific
version of gcc which caused it) and found one open firefox bug
showing the exact problem.  But a lot of people build firefox on, or
for, old systems using i686 or x86_64 (e.g. distros do it!) and
nobody else had seen the problem.

ĸen
-- 
Nanny Ogg usually went to bed early. After all, she was an old lady.
Sometimes she went to bed as early as 6 a.m.


More information about the blfs-dev mailing list