234627 - Mozilla won't start

Status: RESOLVED INVALID

(SeaMonkey :: General, defect)

Description

[Shanmu]

User-Agent:       
Build Identifier: Mozilla/5.0 (X11; U; OSF1 alpha; en-US; rv:1.6) Gecko/20040126

Mozilla doesn't come up. This started happening from last week. 


Reproducible: Always
Steps to Reproduce:
1.run mozilla
2.
3.

Actual Results:  
Nothing happens.

Expected Results:  
Mozilla should have started

Comment 1

[Shanmu]

Attachment: output after setting NSPR_LOG_MODULES=all:5.

I am attchaing the log that I got by setting NSPR_LOG_MODULES=all:5.

Comment 2

[Jim Dunn]

Does it core?  Or just not start?  Any chance it is 
similar to bug 231104 that Phil found in Jan?

Unfortunately I can't process the nspr log file, I am
away and have limited net access.  It would probably
be more useful to just post either the last couple
of offending lines or the lines that appear to be
the error (i.e. compare a good nspr log output
with one from this build using "mozilla about:blank").

And as always make sure you try running from a new profile.

Comment 3

[chris hofmann]

would need a patch to block 1.7a at this point...

Comment 4

[Roland Mainz]

chris hofmann wrote:
> would need a patch to block 1.7a at this point..

I wish we could avoid it that this platform is broken every 2nd milestone... ;-(

Comment 5

[Shanmu]

Attachment: Unusual logs.

Comment 6

[Shanmu]

It doesn't core. it doesn't just start.  The last few offending lines are pasted
here. Afetr printing this stuff it never comes up.  

-1071967584[1400348c0]: WARNING: Cannot create startup observer :
service,@mozilla.org/typeaheadfind;1, file nsAppStartupNotifier.cpp, line 113
WARNING: Cannot create startup observer : service,@mozilla.org/typeaheadfind;1,
file nsAppStartupNotifier.cpp, line 113
-1071967584[1400348c0]: WARNING: dependent window created without a parent, file
nsWindowCreator.cpp, line 128
WARNING: dependent window created without a parent, file nsWindowCreator.cpp,
line 128
GFX: dpi=116 t2p=0.0833333 p2t=12 depth=24
we don't handle eBorderStyle_close yet... please fix me
++WEBSHELL == 1
++DOMWINDOW == 1

Comment 7

[Asa Dotzler [:asa]]

without a DEC champion stepping forward to look into this, not going to block
the release of beta.

Comment 8

[Andrew Schultz]

is this still a problem?  it looks like loading up typeaheadfind failed.  have
you tried clobbering extensions/typeaheadfind (or the whole tree) and rebuilding?

Comment 9

[Shanmu]

I had some success lately in tracking down the problem. When I removed the
--enable-svg from .mozconfig script and built mozilla, I am able to bring the
browser up.  I am working towards a fix for this. 

Comment 10

[Shanmu]

I think this is not a blocker anymore. 

Comment 11

[Roland Mainz]

Alex:
Can you take a look at this issue, please ?

Comment 12

[Alex Fritze]

I don't see how the SVG code could lead to typeaheadfind not starting.
Can you paste your .mozconfig, please?
Which SVG rendering engine (gdi+,libart,cairo) are you using?
Have you tried a clobber build? 

Comment 13

[Shanmu]

Well, when I enabled SVG the browser itself is not coming up. 
I was using libart engine when I foud this problem.
I removed the mozilla directory and pulled all over again, but
still saw that mozilla not coming up. But, as I said earlier,
I get the browser up when I removed the following lines from my .mozconfig

ac_add_options --enable-svg
mk_add_options MOZ_INTERNAL_LIBART_LGPL=1
MOZ_INTERNAL_LIBART_LGPL=1


The .mozconfig that gets me the browser working is 

ac_add_options --enable-tests
ac_add_options --enable-debug
ac_add_options --disable-optimize
ac_add_options --enable-crypto

Comment 14

[Alex Fritze]

> ac_add_options --enable-svg
> mk_add_options MOZ_INTERNAL_LIBART_LGPL=1
> MOZ_INTERNAL_LIBART_LGPL=1

These options are not sufficient to build SVG support. See the SVG build
instructions at http://www.mozilla.org/projects/svg/build.html.
(You need to specify a renderering engine, e.g. "ac_add_options
--enable-svg-renderer-libart")

If the renderer option is omitted, your build should stop with an error during
the compilation of the SVG code. There is a bug (#234185) on adding a check that
a renderer has been defined at configure time.