Re: _OPSYS_HAS_OSSAUDIO

[nia <nia%NetBSD.org@localhost>]

On Mon, Apr 07, 2025 at 08:52:13AM -0700, Alistair Crooks wrote:
> I fail to see what this would gain you.
> 
> If the settings of the definition (the yes/no settings) are wrong, then
> they should be fixed
> 
> If the definition's name is misleading, then it should be fixed
> 

Both are wrong, and this variable being mostly unused (except by one
obscure package which is violating namespace rules) means it's gone
unnoticed.

> The whole idea of abstracting away from os tests is to make this scalable
> and workable. Even/especially for corner cases. I can assure you that with
> the number of platforms that pkgsrc can help, this is the only way to do
> it, and this proposal concerns me because it fundamentally doesn't
> understand the issues

Of course, I'm not advocating replacing this variable with OS-specific
tests.  OSS_TYPE should be checked against "none" instead. This is
reliable and is expected to work properly, and is already used all
over the tree.  It is also completely unambigous.