Bug 235344 – x11-libs/libview-0.6.1 and dev-cpp/libsexymm-0.1.9 has some automagical dependency on USE=accessibility in dev-cpp/gtkmm which causes undefined symbol: _ZThn12_N4view10FieldEntry17delete_text_vfuncEii in vmware-player

• Status: RESOLVED FIXED
• Product: Gentoo Linux

[Description Vlastimil Babka (Caster) (RETIRED) 2008-08-21 07:29:00 UTC]

Tried to emerge vmware-player-2.0.4.93057 (using vmware-overlay but I think it should not be the cause of the problem)

     Wed Aug 20 23:20:33 2008 >>> dev-cpp/gtkmm-2.12.7
     Wed Aug 20 23:24:25 2008 >>> x11-libs/libview-0.6.1
     Wed Aug 20 23:26:11 2008 >>> dev-cpp/libsexymm-0.1.9
     Wed Aug 20 23:28:44 2008 >>> dev-cpp/libgnomecanvasmm-2.22.0

Here vmware-player bailed out in pkg_setup saying that I should rebuild gtkmm with accessibility flag. OK...

     Wed Aug 20 23:45:58 2008 >>> dev-cpp/gtkmm-2.12.7
     Wed Aug 20 23:59:23 2008 >>> app-emulation/vmware-player-2.0.4.93057

Tried to run vmplayer:
/opt/vmware/player/lib/bin/vmplayer: symbol lookup error: /opt/vmware/player/lib/lib/libvmwareui.so.0/libvmwareui.so.0: undefined symbol: _ZThn12_N4view10FieldEntry17delete_text_vfuncEii

Googled, found a possible way to fix in the forum: http://forums.gentoo.org/viewtopic-p-5000045.html?sid=72edbb84e2eac52d08cb2b6854ad1d3f
Which says to remerge these two:

     Thu Aug 21 00:13:53 2008 >>> dev-cpp/libsexymm-0.1.9

Didn't help yet.

     Thu Aug 21 00:19:45 2008 >>> x11-libs/libview-0.6.1

Now it works.
Which means that libview builds some extra symbols based on the accessibility flag in gtkmm? I think in that case it should also introduce the flag and depend on gtkmm's flag? At least the "rebuild gtkmm with accessibility" message in vmware-player should add "and rebuild libview after that"...

[Comment 1 David CHANIAL 2008-08-23 08:58:06 UTC]

I confirm.

i have emerged vmplayer saying me i have to rebuild gtkmm with accessibility use flag.

i rebuild gtkmm with accessibility use flag

vmplayer now build

vmplayer not starting : error :
[...] undefined symbol: _ZThn12_N4view10FieldEntry17delete_text_vfuncEii [...]

so i have rebuild libview

vmplayer start :)

Thanks Vlastimil

[Comment 2 Steve Kutnar 2008-09-28 15:40:01 UTC]

I ran into this with vmware-workstation-6.0.5.109488 as well.  The same libraries are required.

[Comment 3 John-Michael Fischer 2008-10-28 06:20:43 UTC]

patch just worked for me as well.  Encountered when attempting to emerge vmware-workstation and specifically the libview dep.

[Comment 4 Mike Auty (RETIRED) 2008-11-17 10:32:30 UTC]

*** Bug 246642 has been marked as a duplicate of this bug. ***

[Comment 5 Mike Auty (RETIRED) 2008-11-17 11:04:41 UTC]

I'd like the gnome-mm people's input on this, since it either means trying to patch libview to add an accessibility flag (which should fail if gtkmm doesn't have accessibility), or simply fixing on the accessibility USE flag on gtkmm permanently.  It's not clear what the best solution is, although there was a discussion on IRC.

It would be nice to find a solution since I'm beginning to get more dupes turning up with symbol errors (or worse, people using the VMWARE_USE_SHIPPED_GTK=force workaround)...

Also, I'd like to again ask if it's ok to either maintain the package or at least add vmware as the herd on libview?  We seem to be the only packages that make use of it, so it's a bit odd not to receive bug reports about it...  5;)

I'll attach the IRC conversation for those that missed it...

[Comment 6 Mike Auty (RETIRED) 2008-11-17 11:07:40 UTC]

Created attachment 172076 [details]
Conversation from IRC concerning gtkmm/libview/vmware and USE="accessibility"

Ok, very hastily editted, sorry if I cut anything relevant.  It isn't entirely clear what came out of the conversation, but there did seem to be the thought that having gtkmm's ABI change with and without accessibility was a bad thing, and perhaps the best solution might be to remove USE="accessibility" from gtkmm and treat it as-if always set.

In that case, should a different bug be filed, or should this one be reassigned, or is that not the case?  5:)

[Comment 7 Daniel Gryniewicz (RETIRED) 2008-11-17 16:50:20 UTC]

Seeing as atk is an unconditional dep of gtk which is an unconditional dep of gtkmm, I'd say we should just ditch the accessibility flag for gtkmm and have it always on.  Probably a different bug is in order.

[Comment 8 Michael C. Vergallen 2008-11-20 08:00:15 UTC]

Now I get this :

 Attempt to remove filter function 0x7f36ba5229b0 user data 0xacc0e0, but no such filter has been added D-Bus not built with -rdynamic so unable to print a backtrace


How do I fix this now ? 

[Comment 9 Mike Auty (RETIRED) 2008-11-20 10:49:00 UTC]

That quote seems completely out of context.  It doesn't sound related to this bug, so if you really meant to post it on the end of this one, please open a new bug and give us all the details of what happened, what you were doing, and where you saw this error message.  Thanks...  5:)  

[Comment 10 Michael C. Vergallen 2008-11-20 10:58:20 UTC]

(In reply to comment #9)
> That quote seems completely out of context.  It doesn't sound related to this
> bug, so if you really meant to post it on the end of this one, please open a
> new bug and give us all the details of what happened, what you were doing, and
> where you saw this error message.  Thanks...  5:)  
> 
It seemed like it was related at first because it appeared after I rebuilt those 3  libs ...but after further inspection you are right this is a new bug .. sorry for the confusion ...

[Comment 11 Mike Auty (RETIRED) 2008-11-26 20:23:10 UTC]

*** Bug 248245 has been marked as a duplicate of this bug. ***

[Comment 12 Mike Auty (RETIRED) 2008-11-27 19:44:09 UTC]

This issue also occurs in similar packages (specifically libsexymm).  See the following duplicate...

[Comment 13 Mike Auty (RETIRED) 2008-11-27 19:45:34 UTC]

*** Bug 248836 has been marked as a duplicate of this bug. ***

[Comment 14 John M. Drescher 2009-01-13 04:17:33 UTC]

Just got this with app-emulation/vmware-player-2.5.1.126130

reemerging 

dev-cpp/libsexymm and 11-libs/libview 

seems to have got me past this bug.

[Comment 15 Mike Auty (RETIRED) 2009-01-15 23:05:21 UTC]

Reassigning since it appears compnerd's retired (see bug 71985).

Any news from the gnome guys on ditching the atk/accessibility USE flags given comment 7?

[Comment 16 Miran Bozicevic 2009-02-22 20:18:51 UTC]

VMWARE_USE_SHIPPED_GTK=yes

Might do the trick.  Kudos
http://princ3.wordpress.com/2007/05/30/vmware-workstation-60-on-gentoo/

[Comment 17 Mike Auty (RETIRED) 2009-03-24 12:20:17 UTC]

Ok, gtkmm now will always be built with the atkmm API, which means that after gtkmm is rebuilt, and then these are rebuilt, the problem will go away.  I'm therefore going to mark this as FIXED.  Be aware that once gtkmm is rebuilt from USE="-accessibility", things may break, so run a revdep-rebuild if you notice anything breaking...