Re: Combining NOT_FOR_* and BROKEN_ON_*

[David Holland <dholland-pkgtech%netbsd.org@localhost>]

On Wed, Apr 02, 2025 at 03:12:51PM +0200, Thomas Klausner wrote:
 > We currently have two settings to disable packages for particular
 > platforms:
 > 
 > NOT_FOR_* for packages where it "doesn't make sense" for them to be
 > built on a platform, and BROKEN_ON_* for packages where it does, but
 > they don't build for some reason.
 > 
 > I understand the difference between the two, but I don't see it as a
 > useful difference to make - the result is the same, we don't try
 > building on that platform.

That is because, to be blunt, Joerg refused to support it in pbulk and
nobody else has gotten around to it.

In an ideal world the distinction would affect the pbulk reporting so
we don't have to wade through "failure" reports about packages that
are NOT_FOR the target. That would then lead to more attention to the
real failures.

Same for PKG_SKIP_REASON and PKG_FAIL_REASON, which are the underlying
logic.

I swear this point came up only a few days ago.

Anyway, I'm quite against any such change. Far better to get around to
making it work properly.

 > I don't remember a single commit (but there must have been some) where
 > a BROKEN_ON package was fixed, but I remember at least three recent
 > discussions about which of the two variables should be set.

I think that's because we've always had a bias against marking things
BROKEN (rather than either fixing them or deciding it's time to remove
them) and a tendency for people to not understand the distinction and
use BROKEN_ON for things that should be skipped entirely.

Maybe we should deprecate both and replace them with PKG_SKIP_ON and
PKG_FAIL_ON.

(and maybe replace uses of BROKEN with PKG_FAIL_REASON?)

-- 
David A. Holland
dholland%netbsd.org@localhost