Mike Hommey [:glandium] Assignee
	Description • 10 years ago

The error occurred while processing the following file or one of the files it includes:

    /tmp/build/memory/mozalloc/tests/moz.build

The error occurred when validating the result of the execution. The reported error is:

    USE_LIBS contains "nspr", which does not match any LIBRARY_NAME in the tree.

	Dwayne Bailey
	Comment 1 • 10 years ago

This impacts building language packs on http://mozilla.locamotion.org/projects/firefox --disable-compile-environment allows us ot build quicker and without all the other full build junk on our server.  Plus it take a few less resources as we're building mostly every hour.

	Florian Quèze [:florian]
	Comment 2 • 9 years ago

The attached patch let me run configure successfully with --disable-compile-environment, but it doesn't feel great.

Would there be a way to treat the USE_LIBS and HOST_USE_LIBS variables as empty throughout the tree when CONFIG['COMPILE_ENVIRONMENT'] is false?

	Michael Shal [:mshal]
	Comment 3 • 9 years ago

(In reply to Florian Quèze [:florian] [:flo] (Away until March 10) from comment #2)
> Would there be a way to treat the USE_LIBS and HOST_USE_LIBS variables as
> empty throughout the tree when CONFIG['COMPILE_ENVIRONMENT'] is false?

If so, I think a similar approach could be used for other variables, such as SDK_BINARY (see: https://bugzilla.mozilla.org/show_bug.cgi?id=1135684#c2 )

	Mike Hommey [:glandium] Assignee
	Comment 4 • 9 years ago

My needinfo answer will come in the form of patches.

	Mike Hommey [:glandium] Assignee
	Comment 6 • 9 years ago

This allows me to go through a clobber --disable-compile-environment build in 1 minute. Only tested on Linux. Other platforms can go through followups.

	Mike Hommey [:glandium] Assignee
	Comment 7 • 9 years ago

This version passes tests without modifying them.

	Ryan VanderMeulen [:RyanVM]
	Comment 8 • 9 years ago

https://hg.mozilla.org/mozilla-central/rev/84f6168fbf03
https://hg.mozilla.org/mozilla-central/rev/98d0c08747c1

	Stoyan Dimitrov [:stoyan]
	Comment 9 • 9 years ago

Will this be uplifted to mozilla-aurora, mozilla-beta?

	Mike Hommey [:glandium] Assignee
	Comment 10 • 9 years ago

I don't know. Is there a reason to?

	Dwayne Bailey
	Comment 11 • 9 years ago

(In reply to Mike Hommey [:glandium] from comment #10)
> I don't know. Is there a reason to?

I need it in Aurora for automated language pack building on Pootle where we focus only on Aurora.  Having to have the complete compile environment is risky and time consuming.  Having said that if this lands in Aurora 39 that's OK with me as we'll be able to pick it up then.

	Stoyan Dimitrov [:stoyan]
	Comment 12 • 9 years ago

(In reply to Mike Hommey [:glandium] from comment #10)
> I don't know. Is there a reason to?

The guide to manually build language packs (in MDN) has a warning pointing to this bug, so maybe not an issue for many of the localizers.

Dwayne, are those language packs accessible for the mere localizers or are for internal use only?

	Dwayne Bailey
	Comment 13 • 9 years ago

(In reply to Стоян Димитров [:stoyan] from comment #12)
> (In reply to Mike Hommey [:glandium] from comment #10)
> > I don't know. Is there a reason to?
> 
> The guide to manually build language packs (in MDN) has a warning pointing
> to this bug, so maybe not an issue for many of the localizers.

The localisers on Pootle are our deep unskilled localisers, they won't be building language packs themselves, ever...

> Dwayne, are those language packs accessible for the mere localizers or are
> for internal use only?

They're accessible to the localiser, not easily accessible though e.g. http://mozilla.locamotion.org/export/POOTLE_EXPORT/firefox/sco/

They're for debugging purposes for the localiser and only useful for those who aren't yet in Mercurial where nightly builds take over the responsibility.

So if its in Aurora 39 I'm OK with that.  The packs build fine now, just with risks and resources.

	Mike Hommey [:glandium] Assignee
	Comment 14 • 9 years ago

FWIW, it has been broken for more than 6 months, I guess waiting a couple more weeks is not going to make much difference :)

	Dwayne Bailey
	Comment 15 • 9 years ago

(In reply to Mike Hommey [:glandium] from comment #14)
> FWIW, it has been broken for more than 6 months, I guess waiting a couple
> more weeks is not going to make much difference :)

Indeed, but been whining for about as long :)