Re: Combining NOT_FOR_* and BROKEN_ON_*

[Greg Troxel <gdt%lexort.com@localhost>]

Thomas Klausner <wiz%NetBSD.org@localhost> writes:

1> Hi!
>
> 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.

It documents for the reader whether there is something to be done
(BROKEN_ON), or whether there is not (NOT_FOR).

> It also has lead to a couple discussions already if one or the other
> is more appropriate for a particular package, none of which I remember
> as being useful.

Those would be avoided by people using BROKEN_ON in the first place :-)
And, if simply fixed, would be like a vast number of "this isn't quite
right" comments.  (I make mistakes all the time, and sometimes people
just fix them, and sometimes people tell me, and I try to fix them
quickly.)

There seems to be a culture of "upstream is deliberately indifferent to
this platform (or to any computer older than 5 years old that lacks 16
GB of RAM, or whatever)" beyond pkgsrc, and within pkgsrc "mark it
NOT_FOR, accepting upstream's indifference as valid".

> To avoid the not very useful discussions, I suggest removing either of
> them; probably BROKEN_ON_*, but I don't really care.

I remember an earlier comment that BROKEN_ON was bad, because it hid the
failure and lacked any detection that a package might have been fixed.
This objection did not apply to NOT_FOR, more or less.  However, we
don't seem to have anybody doing bulk builds ignoring BROKEN_ON, so that
may be a point without a purpose.


All in all, I don't think it's a good idea to merge these, but I also
can't say that it will matter much in the end.