Re: Combining NOT_FOR_* and BROKEN_ON_*

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

Edgar Fuß <ef%math.uni-bonn.de@localhost> writes:

>> and one is in my opinion completely unreasonable (there is some build issue 
>> and so I'm going to mark it so that nobody can see what the build issue is 
>> any more, thus making it harder for someone to actually fix,
> What do you suggest for the case a package used to build/work on NetBSD, 
> but now doesn't any more because upstream switched to relying on some 
> undefined behaviour being the Linux way?
> So, in essence, "upstream broke it and we can't reasonably fix it".

As documented, that's clearly BROKEN_ON.  Upstream should be written to
POSIX, and isn't, and needs a bugfix.

NOT_FOR should be extremely rare, when there is a valid reason why
excluding the platform is fundamentally the right thing.  The canonical
example makes more sense in terms of ONLY_FOR, when the purpose of the
program is to use a non-standard API that is present only on a small
number of systems.