Add proper logic for debuginfo enablement

[pmatilai]

All these years, enabling debuginfo has required distros to hijack the spec %install section with a macro like this:

%install %{?_enable_debug_packages:%{?buildsubdir:%{debug_package}}}\
%%install\
%{nil}

This for a widely used, longtime upstream supported feature is just gross, and also very non-obvious, feeble and whatnot. And totally prevents the new append/prepend options from being used with %install.

Take advantage of several newish features to make this happen: we need expressions to properly handle the numeric %_enable_debug_packages value from a macro, and if enabled, output the debuginfo template as a dynamic .specpart.

Enable debuginfo on Linux by default in the platform configuration. Notably noarch should not have debuginfo so it's disabled in the platform configuration - since 96467dc we can now actually rely on the platform configuration being valid, so we can drop the "%ifnarch noarch" from the debug package check. Further streamlining should be possible.

specstep.spec needs to be made noarch here, otherwise it'll now try to produce debuginfo and fail.

Fixes: #2204
Fixes: #1878

[pmatilai]

The first two commits are simply #3084 which is needed for this to work.

[pmatilai]

It'd probably be possible to further drop the __debug_package logic and merge __debug_install_post into this, and maybe this should be in couple of separate commits, just wanted to get this out before the weekend 😅

[Conan-Kudo]

This is a lot cleaner and I like this setup.

[pmatilai]

Yeah it's nice to be able to use the dynamic spec stuff for our own purposes.

To clarify, #3084 is only needed for the "rpmbuild %caps" test which intentionally does not use %setup to test that case, and whose debuginfo now gets generated. So that's basically another bug fixed / limitation eliminated, I could've sworn we have a ticket on debuginfo requiring %setup but can't find it.

I haven't tested but it may well cure #3042 too because the combo means it no longer requires the full %setup thing to run.

[pmatilai]

Mentioned the %setup-less debuginfo case and added an explicit sub-test for it.

[pmatilai]

Had a brief look at killing %__debug_package, but that gets more complicated than it should (and who's surprised?) So many packages rely on redefining %debug_package to %{nil} for disabling debuginfo generation that we just have to preserve that for the time being, and having %debug_package define something else as a side-effect is an easy (if ugly) way to handle this. Not worth messing with just now.

[ffesti]

This is even cleaner than my own variant. Great so see we got this to the point where it can be done this cleanly.

[ffesti]

Release notes need to mention this and that distribution need to drop the

%install %{?_enable_debug_packages:%{?buildsubdir:%{debug_package}}}\
%%install\
%{nil}

hack in *-rpm-config. So this probably needs to go into the "Compatibility" section at least in parts.

[pmatilai]

Just noting it here that since distros are widely overriding %__spec_install_post, they'll need to update that to match the new logic in there. @ffesti's version placed the %_enable_debuginfo_packages test into %__spec_install_pre which isn't as widely overridden and so would be more backwards compatible, but then it wouldn't be all in one place. It's a hugely annoying mess 😒

[dmnks]

Just noting it here that since distros are widely overriding %__spec_install_post, they'll need to update that ...

I suppose this was obsoleted by #3098, correct?

[pmatilai]

Oh, yup. I thought I made a further comment here but apparently didn't...