Bugzilla Bug 46568

Compile aborts with following error:

source='AudioEngineInfoForm_UI.cpp' object='AudioEngineInfoForm_UI.o' libtool=no \
depfile='.deps/AudioEngineInfoForm_UI.Po' tmpdepfile='.deps/AudioEngineInfoForm_UI.TPo' \
depmode=gcc3 /bin/sh ../admin/depcomp \
g++ -DHAVE_CONFIG_H -I. -I. -I.. -I../lib -I/usr/qt/3/include -I/usr/X11R6/include   -DQT_THREAD_SUPPORT  -D_REENTRANT  -O3  -c -o AudioEngineInfoForm_UI.o `test -f 'AudioEngineInfoForm_UI.cpp' || echo './'`AudioEngineInfoForm_UI.cpp
AudioEngineInfoForm_UI.cpp: In constructor `
   AudioEngineInfoForm_UI::AudioEngineInfoForm_UI(QWidget*, const char*,
   unsigned int)':
AudioEngineInfoForm_UI.cpp:75: error: `Spacer2' undeclared (first use this
   function)
AudioEngineInfoForm_UI.cpp:75: error: (Each undeclared identifier is reported
   only once for each function it appears in.)
AudioEngineInfoForm_UI.cpp:80: error: `Spacer1' undeclared (first use this
   function)
make[3]: *** [AudioEngineInfoForm_UI.o] Fehler 1
make[3]: Leaving directory `/var/tmp/portage/hydrogen-0.8.2/work/hydrogen-0.8.2/hydrogen_gui'
make[2]: *** [all-recursive] Fehler 1
make[2]: Leaving directory `/var/tmp/portage/hydrogen-0.8.2/work/hydrogen-0.8.2/hydrogen_gui'
make[1]: *** [all-recursive] Fehler 1
make[1]: Leaving directory `/var/tmp/portage/hydrogen-0.8.2/work/hydrogen-0.8.2'
make: *** [all] Fehler 2

!!! ERROR: media-sound/hydrogen-0.8.2 failed.
!!! Function src_compile, Line 72, Exitcode 2
!!! (no error message)

After looking through the code the reason is two missing QSpacerItem type member declarations from the header. This also applies to LadspaFXSelector_UI.h and PreferencesDialog_UI.h

Created an attachment (id=28549) [details]
Fixes missing declarations from the hydrogen headers.

To be placed into the files/ dir.

Created an attachment (id=28550) [details]
Adds epatch the ui-spacer patch and the missing addwrite call

This includes the sandbox violation fix for bug #40347.

torbenh, you're listed as maintainer in this package, so I'll hand this off to
you.  I removed some old ebuilds and added the qt addwrite to the remaining
ones, but you should check out this patch to see if this is how you want to
handle it.

Michael, your patch to the ebuild (comment #2) contains a "#" before epatch,
so nothing is patched - but after removing this "#" it works really fine.

Thank you.

Oops, yes, I commented out the epatch line to reconstruct the exact error
message for filing this bug, and accidently diffed agains that ebuild ;-)

Hmm, this is very strange..

When patching the unpacked directory "by hand" and then running configure; make install, everything compiles ok.. but when i let the ebuild patch the sources, it's getting patched, but still runs into the same compile errors as if it wasn't (and yes, i *have* uncommented the epatch line.. =)

the unpacked sources in the /var/tmp/portage/ directory *really* get patched (i even diffed the two directories, i didn't trust emerge saying it was patching it when i saw that it still produced the same error... but no, they're identical %\

can anyone reproduce this, or do i just need some sleep..?

I'm also having the problem, its not just you.

I just cleaned all and emerged again to check, and for me it worked without
problems (using portage-2.0.50-r6).

But I've found a related description of a "manual configure works, but econf
does not"-Phenomenon at http://bugs.gentoo.org/show_bug.cgi?id=47329#c7 

So I assume it's a problem with portage. Perhaps you should file a separate bug
on this.

setting WORKSFORME

I've got the same problem, on a clean system - same error message as the first
comment.
ONLY With the patch it compiles fine.

What's the deal with this patch not being added to the ebuild?  Is there
something wrong with it?  The patch works for me, and hydrogen was broken
without it.

Re-opening, I just hit the same compilation failure.
Running w/ audiofile-0.2.5, qt-3.3.0-r1, gcc 3.3.2-r5.  
Fails regardless of use flags, 

torbenh is non-responsive... reassigning...

Jeremy, thoughts on the addwrite?  iirc, I'd thought you were futzing w/ a
similiar issue...
Aside from that, looking at the upstream code (coz I'm lazy), they threw out
the Spacer* from the ui definition, between ver 1.12 and 1.15.

http://cvs.sourceforge.net/viewcvs.py/hydrogen/hydrogen/hydrogen_gui/AudioEngineInfoForm_UI.ui?r1=text&tr1=1.13&r2=text&tr2=1.12&diff_format=h

I'd be inclined to do the same, although it is varying from the original a
bit..  Thoughts?
I'd rather chuck the spacer* definitions, since they're not even used, although
it ought to change the layout a bit. I don't use hydrogen, although I can
emerge it for testing...

the addwrite gets set when you use 'inherit kde' and use the kde_src_install (i
believe that's correct)... or youcan do it manually...

as for the spacer stuff... I don't really use this app, so I'll trust your
judgement =)

*** Bug 51056 has been marked as a duplicate of this bug. ***

I do use the hydrogen drummachine (although from the cvs release); I can bake
and test a compilation patch if you wish.

Committed to CVS as hydrogen-0.8.2-r1. thanks for the patch!

I just had this problem because the unpatched ebuild is still the one available
by default in portage. Using "ACCEPT_KEYWORDS=~x86 emerge hydrogen" I was able
to compile hydrogen-0.8.2-r4. Since the actual ebuild(hydrogen-0.8.2.ebuild)
does not work should not it be removed in favor of a working one(r4 is fine
here) before saying the problem is fixed?