• Use a comma separated list of subpkg
• Add documentation for TARGET_SUBPACKAGES variable

In general, it should works, but the *DEPENDS_ALL needs to stay (even if not used by the framework) as they are used externally (by poudriere).

It would be preferable to have pkg-descr.gtk3 and pkg-descr.pango to provide an appropriate description for the subpackages.

Don´t you need an additional pkgdescr.basu?

It looks good to me.

The idea is that every time a USES is included, it generates a USES_*_DEPENDS
Those prefixed depends are going to be copied to the appropriate *_DEPENDS (without suffix main package, with suffixes for the subpackages)

It's a valid point.
I guess we need to work on USES/USE_ support heavily.

I understand what you mean.
If I understand correctly what you are trying to do, you define the glx subpkg that is built iff the GLX option is enabled.
A solution could be to not use GLX_LIB_DEPENDS.glx but only LIB_DEPENDS.glx
If the subpackage glx is not built, the dependency should be ignored.

Hi. I'm a bit confused by your comments.
If a subpackage is associated to an OPTION, what is the need to duplicate everything with <opt>_*_DEPENDS.subpkg ?
There is no need for subpkg to be there.
subpkg is a way to create multiple packages from the same build, not to configure builds.
Am I missing something?

LGTM

• Fix users of _LICENSE_DIR

• Fix multi/dual license

• Initial import of the subpackage framework
• Fix [build|run]-depends-list targets
• Fix type in PKGFILE
• Fix METADIR
• Fix type in RE
• Fix PLIST
• Fix _PKGDIR initialization
• Fix typo
• Fix the case when there are no subpackages
• Fix _PLIST
• Simplify METADIR as in the original patch
• Fix _strip_perms
• More fixes
• Fix package creation
• Add SUBPACKES to the list of OPTIONS_HELPERS
• Make DESCR.subpkg warning message shorter
• Fix PLIST_FILES.subpkg
• Implement LICENSE support
• Fix dependencies in sub-packages
• Fix actual-package-depends target
• Add PKGBASE.subpkg variable
• Make do-depends aware of the subpackages
• do-depends.sh: fix a typo
• Add annotation to packages about being a subpkg
• Remove a weird copy&pasta leftover
• Fix subpackages in some dependency list

rebase

• Remove a weird copy&pasta leftover

I understand.
In this case, my only observation was that we are using a case/esac construct instead of an if/fi, but the matching condition seems more natural with case/esac.

It seems you removed the case when file is not */*, but matching *.
Is it always the case that file matches */*?
If yes, why even test it? If not, shouldn't we manage it as it was before, without creating the directory?

• Trying to fix the build of subpkgtest2

• Add annotation to packages about being a subpkg

Add subpackage dependency as test

• Initial import of the subpackage framework
• Fix [build|run]-depends-list targets
• Fix type in PKGFILE
• Fix METADIR
• Fix type in RE
• Fix PLIST
• Fix _PKGDIR initialization
• Fix typo
• Fix the case when there are no subpackages
• Fix _PLIST
• Simplify METADIR as in the original patch
• Fix _strip_perms
• More fixes
• Fix package creation
• Add SUBPACKES to the list of OPTIONS_HELPERS
• Make DESCR.subpkg warning message shorter
• Fix PLIST_FILES.subpkg
• Implement LICENSE support
• Fix dependencies in sub-packages
• Fix actual-package-depends target
• Add PKGBASE.subpkg variable
• Make do-depends aware of the subpackages
• do-depends.sh: fix a typo

Rebase

• Fix dependencies in sub-packages

• Fix English in the error message