#1 Rebuild for libinstpatch 1.1.6 (and fix FTBFS)
Merged 3 years ago by kevin. Opened 3 years ago by music.
rpms/ music/muse libinstpatch-1.1.6  into  rawhide

@@ -0,0 +1,12 @@ 

+ diff -Naur muse-3.0.2-original/CMakeLists.txt muse-3.0.2/CMakeLists.txt

+ --- muse-3.0.2-original/CMakeLists.txt	2018-01-29 16:10:59.000000000 -0500

+ +++ muse-3.0.2/CMakeLists.txt	2021-04-17 08:47:47.027195203 -0400

+ @@ -441,7 +441,7 @@

+      ##

+      

+      if(ENABLE_LV2_GTK2)

+ -      find_package(GTK2 COMPONENTS gtk gtkmm)

+ +      PKG_CHECK_MODULES(GTK2 REQUIRED gtkmm-2.4)

+        if(GTK2_FOUND)

+            set(HAVE_GTK2 ON)

+        endif(GTK2_FOUND)

file modified
+13 -2
@@ -11,7 +11,7 @@ 

  # See: https://fedoraproject.org/wiki/AudioCreation

  Epoch:         1

  Version:       3.0.2

- Release:       14%{?dist}

+ Release:       15%{?dist}

  # original freeverb plugin was public domain

  # givertcap (not built) is GPLv2

  # The rest, including the core of muse is distributed under GPLv2+
@@ -27,6 +27,10 @@ 

  # Qt include fixes for newer Qt. From upstream trunk

  # https://github.com/muse-sequencer/muse/commit/99816883

  Patch2:        muse-qt-include-fixes.patch

+ # Find gtkmm using pkg_check_modules() so we have the correct compiler and

+ # linker flags for its recursive dependencies, such as atkmm. Fixes FTBFS

+ # RHBZ#1923460.

+ Patch3:        muse-3.0.2-find-gtkmm-with-pkgconfig.patch

  

  BuildRequires: alsa-lib-devel

  BuildRequires: cmake
@@ -49,7 +53,7 @@ 

  BuildRequires: qt5-qttools-static

  BuildRequires: rtaudio-devel

  # For lv2 plugins

- BuildRequires: gtkmm24-devel

+ BuildRequires: pkgconfig(gtkmm-2.4)

  

  Requires:      hicolor-icon-theme

  
@@ -65,6 +69,7 @@ 

  %patch0 -p1 -b .includes

  %patch1 -p1 -b .fluidsynth1and2

  %patch2 -p2 -b .qtdep

+ %patch3 -p1 -b .gtkmm-pkgconf

  

  # Convert to Python3. Only "print" and "raw_input" occurences

  2to3 --write --nobackups utils/muse*
@@ -106,6 +111,12 @@ 

  %{_datadir}/metainfo/%{name}.appdata.xml

  

  %changelog

+ * Sat Apr 17 2021 Benjamin A. Beasley <code@musicinmybrain.net> - 1:3.0.2-15

+ - Rebuild for libinstpatch 1.1.6

+ - Find gtkmm using pkg_check_modules() so we have the correct compiler and

+   linker flags for its recursive dependencies, such as atkmm. Fixes FTBFS

+   RHBZ#1923460.

+ 

  * Tue Jan 26 2021 Fedora Release Engineering <releng@fedoraproject.org> - 1:3.0.2-14

  - Rebuilt for https://fedoraproject.org/wiki/Fedora_34_Mass_Rebuild

  

Please build into the appropriate side tag so I can update muse together with libinstpatch:

fedpkg build --target=f35-build-side-40054

Or, if you want to merge and not build, I should be able to start the build. Please do not rebuild after merging without using the side tag.

I have tested this PR against the new libinstpatch version by scratch-building in the side tag (https://koji.fedoraproject.org/koji/taskinfo?taskID=66126882).

I will announce the so-version bump on the fedora-devel list shortly.

This PR also fixes your open FTBFS bug, https://bugzilla.redhat.com/show_bug.cgi?id=1923460.

Going ahead and merging this.

Pull-Request has been merged by kevin

3 years ago

Please revert. This is not the correct place to fix this.

The atkmm issue needs to be fixed in cmake:
https://bugzilla.redhat.com/show_bug.cgi?id=1944935

I saw the bug you linked. It does seem like it would be easy enough to back out the patch after the problem is fixed correctly in CMake. A temporary workaround allows the libinstpatch update to go through without converting muse’s FTBFS to an FTI.

Anyway, the libinstpatch update including the muse rebuild is in Rawhide now. You or @kevin can back out the patch and bump the release number again, but the package will be FTBFS again and you won’t be able to rebuild. And the old build wouldn’t work with the new libinstpatch so-version anyway.

I’m sorry it seems I didn’t wait long enough for you to give an opinion on the PR. Given where things stand now, how would you like to proceed? Specifically, is there anything I can do?

All good. Let's keep this build until cmake issue is resolved. Then we will remove the patch.
The only eyesore is the messages in the SPEC claim this is a fix. It's not a fix, it's a workaround, and should be marked as such. If you can update the SPEC with that I would appreciate.

OK, I will submit a new PR to do that. Shall I revise both the comment and the changelog message?

Yes, please. Thank you.