363487 - Build fails in postflight_all when running unify

Status: RESOLVED INVALID

(Firefox Build System :: General, defect)

Description

[Chris Cooper [:coop] (he/him)]

The Fx-Trunk build is failing on one of the new Intel xserves in postflight_all when it tries to run unify. Log excerpt:

/builds/tinderbox/Fx-Trunk/Darwin_8.8.4_Depend/mozilla/build/macosx/universal/unify \
  ../build/universal/ppc/dist/firefox/Minefield.app \
  ../build/universal/i386/dist/firefox/Minefield.app \
  ../build/universal/ppc/dist/universal/firefox/Minefield.app
/builds/tinderbox/Fx-Trunk/Darwin_8.8.4_Depend/mozilla/build/macosx/universal/unify: copyIfIdentical: files differ:
  ../build/universal/ppc/dist/firefox/Minefield.app/Contents/MacOS/components/nsSafebrowsingApplication.js,
  ../build/universal/i386/dist/firefox/Minefield.app/Contents/MacOS/components/nsSafebrowsingApplication.js
make[1]: *** [postflight_all] Error 1
make: *** [alldep] Error 2

Full log is here: 
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1165870320.9847.gz&fulltext=1

Comment 1

[J. Paul Reed [:preed]]

Adding the usual suspects.

Comment 2

[J. Paul Reed [:preed]]

Mento:

If you have a couple of free seconds, could you look at this? Would love to get builds from these new (really fast!) machines. :-)

Comment 3

[Mark Mentovai]

I don't think I have access to that tinderbox.  Can someone run diff on the two files it's complaining about?

Comment 4

[Benjamin Smedberg]

Yeah... I suspect the XUL preprocessor adding //@line directives, but we need to know what it's actually adding.

Comment 5

[J. Paul Reed [:preed]]

(In reply to comment #3)
> I don't think I have access to that tinderbox.  Can someone run diff on the two
> files it's complaining about?

Problem might be that those directories don't exist:

bm-xserve08:/builds/tinderbox/Fx-Trunk/Darwin_8.8.4_Depend/build cltbld$ pwd
/builds/tinderbox/Fx-Trunk/Darwin_8.8.4_Depend/build
bm-xserve08:/builds/tinderbox/Fx-Trunk/Darwin_8.8.4_Depend/build cltbld$ diff -u universal/{ppc,i386}/dist/firefox/Minefield.app/Contents/MacOS/components/nsSafebrowsingApplication.js
diff: universal/ppc/dist/firefox/Minefield.app/Contents/MacOS/components/nsSafebrowsingApplication.js: No such file or directory
diff: universal/i386/dist/firefox/Minefield.app/Contents/MacOS/components/nsSafebrowsingApplication.js: No such file or directory

When I search for the relevant paths, I get to the following directory; tab completion wants to give me:

bm-xserve08:/builds/tinderbox/Fx-Trunk/Darwin_8.8.4_Depend/build cltbld$ diff universal/ppc/dist/Minefield.app/Contents/MacOS/components/nsSafebrowsingApplication.js universal/i386/dist/  
bin           host          lib           public        
branding      idl           package       sdk           
chrome-stage  include       private       xpi-stage     

Comment 6

[Mark Mentovai]

Nah, it's just that you probably ran this while the tinderbox was in the middle of another build in the same tree, and it had already blown away "dist" and hadn't yet put those files back in place.

Comment 7

[J. Paul Reed [:preed]]

(In reply to comment #6)
> Nah, it's just that you probably ran this while the tinderbox was in the middle
> of another build in the same tree, and it had already blown away "dist" and
> hadn't yet put those files back in place.

Yeah, coop pointed that out; *however*:

universal/ppc/dist/firefox/Minefield.app/Contents/MacOS/components/nsSafebrowsingApplication.js 

!=

universal/ppc/dist/Minefield.app/Contents/MacOS/components/nsSafebrowsingApplication.js

Is that expected?

Comment 8

[Chris Cooper [:coop] (he/him)]

(In reply to comment #7)
> Is that expected?

Looks like official branding (or lack thereof) is being interpreted differently:

bm-xserve08:/builds/tinderbox/Fx-Trunk/Darwin_8.8.4_Depend cltbld$ diff build/universal/ppc/dist/firefox/Minefield.app/Contents/MacOS/components/nsSafebrowsingApplication.js build/universal/i386/dist/firefox/Minefield.app/Contents/MacOS/components/nsSafebrowsingApplication.js
1921,1922c1921,1922
< //@line 58 "/builds/tinderbox/Fx-Trunk/Darwin_8.8.4_Depend/mozilla/browser/components/safebrowsing/src/../content/globalstore.js"
< const MOZ_OFFICIAL_BUILD = true;
---
> //@line 60 "/builds/tinderbox/Fx-Trunk/Darwin_8.8.4_Depend/mozilla/browser/components/safebrowsing/src/../content/globalstore.js"
> const MOZ_OFFICIAL_BUILD = false;

Comment 9

[J. Paul Reed [:preed]]

(In reply to comment #8)
> (In reply to comment #7)
> > Is that expected?
> 
> Looks like official branding (or lack thereof) is being interpreted
> differently:
> 
> bm-xserve08:/builds/tinderbox/Fx-Trunk/Darwin_8.8.4_Depend cltbld$ diff
> build/universal/ppc/dist/firefox/Minefield.app/Contents/MacOS/components/nsSafebrowsingApplication.js
> build/universal/i386/dist/firefox/Minefield.app/Contents/MacOS/components/nsSafebrowsingApplication.js
> 1921,1922c1921,1922
> < //@line 58
> "/builds/tinderbox/Fx-Trunk/Darwin_8.8.4_Depend/mozilla/browser/components/safebrowsing/src/../content/globalstore.js"
> < const MOZ_OFFICIAL_BUILD = true;
> ---
> > //@line 60 "/builds/tinderbox/Fx-Trunk/Darwin_8.8.4_Depend/mozilla/browser/components/safebrowsing/src/../content/globalstore.js"
> > const MOZ_OFFICIAL_BUILD = false;
> 

Related to the changes for bug 362514?

Comment 10

[Mark Mentovai]

tinderbox should set up the variables that dictate whether or not you're building a release build itself.  Maybe with all that's been going on to get this box up, ppc wound up configured for official, then the build failed at some point but ppc never reconfigured, and during the build in which x86 was initially configured, tinderbox had the official variables off.  Force both to reconfigure?  (Clobber?)

Comment 11

[Chris Cooper [:coop] (he/him)]

Not sure how we got into a state with differential OFFICIAL_BUILD versions, but a clobber has cleaned things up.