We need to remove the exec permission on the eln files to prevent them from
having the same build-id and thereby failing the debuginfo extraction, while at
the same time they must be executable for emacs to use them.
We use filelists for that where we use attr to set their executable bits *after*
the build.
The shell snippet comparing %info_files and ls info was not working anyway and
the check itself was not serving any purpose. Also, not all the contents of that
directory are actually installed as info pages.
emacs-libdir-vs-systemd.patch relies on the systemd package config files being
in the buildroot, which stopped being the case in Fedora >= 37. Thus we need to
depend on systemd explicitly
It should be deprecated now by Cairo rendering with the new native pgtk in 28
See current build.log
eg https://koji.fedoraproject.org/koji/buildinfo?buildID=2001145:
Does Emacs use cairo? yes
:
Does Emacs use -lfreetype? yes
Does Emacs use HarfBuzz? yes
Does Emacs use -lm17n-flt? yes
Does Emacs use -lotf? yes
Does Emacs use -lxft? no
- Enables native compilation which requires libgccjit
- Use upstream AppStream metainfo
- Add %{setarch} to make bootstrap as it builds architecture specific
.eln files.
- Move the GTK+ build at the end to avoid rebuilding during make
install in the %install section.
This rebuild does not compile .eln files as the .elc files from nox
build are present and we don't run make bootstrap here. The
resulting .pdmp file does not load .eln files, so Emacs fails to
find exact path to the .eln files at the run time after
installation. See https://debbugs.gnu.org/52376#89
- Use pdmp files with fingerprints
Keeping the .pdmp files alongside the binaries is not recommended. See
https://debbugs.gnu.org/52488
Applying the patch which creates and loads pdmp files with binary
specific fingerprints
https://git.savannah.gnu.org/cgit/emacs.git/commit/?id=e81f1faca4 This
allows us to put these files into their correct location in libexec.
Signed-off-by: Bhavin Gandhi <bhavin7392@gmail.com>
To decide the default value of ispell-program-name, Emacs looks in turn for a number of different binaries, and uses whichever it finds first. The snag with this approach is that, just because a binary is installed, doesnʼt mean that a dictionary in your language is installed for that binary.
Since https://bugzilla.redhat.com/show_bug.cgi?id=713600 weʼve been patching the search list to look for Hunspell before Aspell, presumably because Hunspell dictionaries were considered more likely to be installed than the Aspell dictionaries.
However, now we have Enchant, which will automatically select a backend based on a configuration preference, skipping over any backends for which youʼre missing a dictionary. This seems to be exactly what we want, so letʼs use it here.
