Page MenuHomeFreeBSD

Mk/Uses/gmake.mk: Permit arguments for adding devel/gmake to RUN_DEPENDS
AbandonedPublic

Authored by jrm on Oct 2 2017, 12:30 AM.
Tags
None
Referenced Files
Unknown Object (File)
Jan 28 2024, 9:18 AM
Unknown Object (File)
Dec 23 2023, 12:12 AM
Unknown Object (File)
Dec 14 2023, 6:32 AM
Unknown Object (File)
Sep 21 2023, 5:32 PM
Unknown Object (File)
Sep 11 2023, 4:46 AM
Unknown Object (File)
Aug 1 2023, 5:39 PM
Unknown Object (File)
May 7 2023, 4:50 AM
Unknown Object (File)
Mar 7 2023, 12:23 AM

Details

Reviewers
None
Group Reviewers
portmgr
O5: Ports Framework(Owns No Changed Paths)
Summary

Some ports (e.g., math/R or editors/emacs) include their own packaging functionality. Some of these ports' internal packages require GNU make to build, so devel/gmake must be added to the port's RUN_DEPENDS.

Diff Detail

Repository
rP FreeBSD ports repository
Lint
No Lint Coverage
Unit
No Test Coverage
Build Status
Buildable 11838
Build 12178: arc lint + arc unit

Event Timeline

Not everybody's going to need those though. Can a note about some packages needing gmake be added to pkg-message for the ports in question?

I am now recalling a conversatation about this topic a few months ago. I tend to agree that these nonessential dependencies could be left to the user to manually install, but I also understand why the maintainers of, e.g, www/npm and about a dozen other ports have added a RUN_DEPENDS on devel/gmake. Users will forget and file reports about failing installations of internal packages. I'm ambivalent. If there is a concensus that gmake shouldn't be a run-time dependency in these situations, I will remove the RUN_DEPENDS=gmake:devel/gmake for math/R and follow your advice for a pkg_message.

Personally I think telling users "Install these extra packages in order for the software to work right" provides a bad experience. If you install something, you expect it work completely. If it doesn't and you have to install something else in order for that to happen, that's unexpected. So I think whatever deps are needed should be there. The contents of this patch itself seem fine to me.

I think the 3 ports that need it are better off adding a RUN_DEPENDS by manually.

Sounds good. Upon further reflection, it might not make sense to encourage these optional dependencies.