387336 - [OS/2] enable mouse pointers in libxul builds

Status: RESOLVED FIXED

(Firefox Build System :: General, defect)

Description

[Walter Meinl]

User-Agent:       Mozilla/5.0 (OS/2; U; Warp 4.5; en-US; rv:1.9a7pre) Gecko/2007070900 Minefield/3.0a7pre
Build Identifier: Mozilla/5.0 (OS/2; U; Warp 4.5; en-US; rv:1.9a7pre) Gecko/2007070900 Minefield/3.0a7pre

Currently mouse pointers except the editing and the arrow pointers don't work in libxul-enabled firefox. The reason is, that the firefox.exe is build shared therefore the resources are not compiled in the exe. But wdgtos2.dll that contains these resources for shared builds isn't available either. Thus, we have to put the pointers in xul.dll.

Reproducible: Always

Steps to Reproduce:
1. build firefox with --enable-libxul
2. run, pointers don't change when hovering over a link or for dnd
3.
Actual Results:  
run ff, pointers don't change when hovering over a link or for dnd

Expected Results:  
should change to finger or show an hourglass when loading a page

Comment 1

[Walter Meinl]

Attachment: patch

As a xul.dll gets build even in static or non-libxul shared builds I ifdefed the entry for the rc, that the resources are only compiled into the big xul.dll. The xulrunos2.rc file was merged from widget.rc and the splashos2.rc files of the other applications. By digging in it I found that we could drop a part of the xulrunner/splashos2.rc, as xulrunner doesn't get built statically anymore (the respective lines in the Makefile.in where removed a while ago, while the splashos2.rc file was never adjusted).

Comment 2

[Peter Weilbacher]

This depends on bug 348533, right?

If one builds FF not on top of XULRunner but just with --enable-libxul then widget also gets merged into xul.dll. Does that need similar adjustments? (Sorry, haven't looked at the involved files yet...)

Comment 3

[Walter Meinl]

(In reply to comment #2)
> This depends on bug 348533, right?
> 

> If one builds FF not on top of XULRunner but just with --enable-libxul then
> widget also gets merged into xul.dll. Does that need similar adjustments?
> (Sorry, haven't looked at the involved files yet...)
> 
The fix is entirely the same, and was meant for libxul-enabled firefox (which is some transition state to get to FF on top of XULRunner). Widget is then a static lib (wdgtos2.lib) that's part of libxul.
Thinking and trying a lot around where would be the right place to put the resources, I'm now almost sure that xul.dll is the correct target, cause any XULRunner application would profit then and we don't have to care about it any longer. XULRunner itself doesn't contain any cursors.
BTW, this is also an adoption of the respective checkin for windows (see bug 323732)

Comment 4

[Walter Meinl]

(In reply to comment #2)
> This depends on bug 348533, right?
> 
Mmh, I wouldn't really say that this bug depends on bug 348533. Somehow bug 348533 blocks this bug as it blocks also the ability to build a libxul-enabled Firefox, which is the default build option now. I made there a suggestion there, how we could overcome the blocking, which would of course result in at least a temporary disabling of the java131 compatibility.

Comment 5

[Peter Weilbacher]

(In reply to comment #4)
> Mmh, I wouldn't really say that this bug depends on bug 348533. 
> Somehow bug 348533 blocks this bug

I don't think there is a difference between these two in Bugzilla.

Will look at the patch once I could try it out.

Comment 6

[Peter Weilbacher]

OK, the patch makes sense to me, but I cannot reproduce the problem. I built XULRunner (without Firefox) from trunk, got the mybrowser demo from <http://benjamin.smedbergs.us/xulrunner/>, edited application.ini to 1.9.* and ran it. All mouse pointers (pointer, finger, text, +/- zoom pointers, etc.) appear fine, no matter if I use a patched xul.dll or the original one.

Comment 7

[Peter Weilbacher]

Oh, and I think we should leave xulrunner/app/splashos2.rc untouched for now. At least as long as similar stuff is still in the Windows' splash.rc file.

Comment 8

[Peter Weilbacher]

Comment on attachment 271461 [details] [diff] [review]
patch

Ah, I should quit other Mozilla apps before testing this, otherwise it seems to get the pointers from shared RAM somehow?!

So I'm happy if the splashos2.rc part is removed.

But I would like confirmation from Ted because I'm so utterly clueless about XULRunner and libxul.

Comment 9

[(not currently active) Ted Mielczarek]

Comment on attachment 271461 [details] [diff] [review]
patch

The Makefile portions look ok, but I obviously don't know anything about OS/2 resources.  :)

Comment 10

[Peter Weilbacher]

Patch checked in, without splashos2 part.