542004 - Removal of dependentlibs.list breaks Prism on OS X

Status: RESOLVED FIXED

(Firefox Build System :: General, defect)

Description

[Matthew Gertner]

When Prism for Firefox (i.e. the extension) is used to create a Prism app on OS X, it uses the XUL runtime from Firefox for the app so that it doesn't have to be duplicated on disk. I do this by setting GRE_HOME to point to Firefox.app/Contents/MacOS.

This worked fine in Firefox 3.5, but in 3.6 the file dependentlibs.list has been removed from the distribution (see https://bugzilla.mozilla.org/show_bug.cgi?id=463605#c13). This prevents our stub (based on the XULRunner stub) from loading the XUL libraries.

I guess this problem is fairly specific to Prism since I doubt anyone else is using the Firefox runtime in this way. If there's a better way to accomplish this that I didn't think of, please let me know. Otherwise I'll post a patch putting that file back into the dist on OS X.

Comment 1

[(not currently active) Ted Mielczarek]

Just patch it back. I didn't realize that removing it would break things, obviously, since it was missing from the Win32 packaging manifest (apparently it's not required there).

You should just have to add it back to package-manifest.in, and remove it from removed-files.in.

Comment 2

[Phil Ringnalda (:philor)]

And not ifdeffed for just OS X - I finally got curious enough about it to ask bsmedberg a while back, and he said we should have been shipping it on all platforms all along (ifdef libxul, back in the days when we would package something that wasn't libxul), but I ignored the part where he said it would be too much trouble to not generate it for non-libxul builds, and got stuck trying to make that work.

Comment 3

[Matthew Gertner]

Attachment: Put dependentlibs.list into the manifest

Comment 4

[(not currently active) Ted Mielczarek]

Comment on attachment 423389 [details] [diff] [review]
Put dependentlibs.list into the manifest

As philor points out, that should probably be inside #ifdef MOZ_LIBXUL.

r=me with that change.

Comment 5

[Phil Ringnalda (:philor)]

By which Ted means "why does philor always confuse things with issues from his silly non-libxul app? you only get into Firefox's package-manifest if you are MOZ_LIBXUL, so there's no need for an ifdef here."

Comment 6

[(not currently active) Ted Mielczarek]

Pushed to m-c:
http://hg.mozilla.org/mozilla-central/rev/61e445047db6

Comment 7

[(not currently active) Ted Mielczarek]

Comment on attachment 423389 [details] [diff] [review]
Put dependentlibs.list into the manifest

We should get this into 3.6.1, it broke Prism and it's an accidental regression.

Comment 8

[Matthew Gertner]

Does target milestone 1.9.3a1 mean this won't get into FF 3.6?

Comment 9

[(not currently active) Ted Mielczarek]

No, it's generally just used to mean "this landed in 1.9.3a1". The patch is up for approval for 1.9.2.

Comment 12

[Greg Lloyd]

Pardon the clarification request, but I just reported this as Bug 557007 wrt FF 3.6.3 (MacOS X) and it is classified as a duplicate. Is this patch still pending acceptance in a future public update? If the patch is included in ff 3.6.3 something else is wrong

Comment 13

[Greg Lloyd]

(In reply to comment #12)
> Pardon the clarification request, but I just reported this as Bug 557007 wrt FF
> 3.6.3 (MacOS X) and it is classified as a duplicate. Is this patch still
> pending acceptance in a future public update? If the patch is included in ff
> 3.6.3 something else is wrong

Sorry "Prism broken" reported as Bug513628 and marked as dup...

Comment 14

[Matthew Gertner]

Yeah, we're waiting for approval so this fix can be included into an official Firefox release. It looks like we're hoping to get approval for Gecko 1.9.2.4 which means it would be in Firefox 3.6.4 (someone please correct me if I'm wrong about that).

Comment 15

[Greg Lloyd]

(In reply to comment #14)
> Yeah, we're waiting for approval so this fix can be included into an official
> Firefox release. It looks like we're hoping to get approval for Gecko 1.9.2.4
> which means it would be in Firefox 3.6.4 (someone please correct me if I'm
> wrong about that).

Thanks! I'll post a brief note on the workaround and fix to the Prism Reviews page where a number of people been frustrated by this problem. Any Prism developer folk chiming in on issues like this occasionally would be helpful. I *really* like using Prism and hate to see it get a bad rep it does not deserve.

Comment 16

[LegNeato]

Comment on attachment 423389 [details] [diff] [review]
Put dependentlibs.list into the manifest

a=LegNeato for 1.9.2.5. Please ONLY land this on mozilla-1.9.2 default, as we
are still working on 1.9.2.4 on the relbranch

Comment 17

[Matthew Gertner]

Attachment: Same patch with user, title, etc. set

I followed the instructions at http://blog.bonardo.net/2010/06/22/so-youre-about-to-use-checkin-needed. I guess whoever is committing this should use this patch.

Comment 18

[(no longer active)]

http://hg.mozilla.org/releases/mozilla-1.9.2/rev/a34de5780b91