Re: configure --with-uuid=bsd fails on NetBSD

Nazir Bilal Yavuz <byavuz81(at)gmail(dot)com> writes:
> I updated my patch. I checked version field in 'uuid_generate_internal'
> function instead of checking it in 'uuid_generate_v1' and
> 'uuid_generate_v1mc' functions, but I have some questions:

Yeah, that seems like the right place. I tweaked the code to check
strbuf not str just so we aren't making unnecessary assumptions about
the length of what is returned. strbuf[14] is guaranteed to exist,
str[14] maybe not.

> 1 - Should it be checked only for  '--with-uuid=bsd' option?
>     1.1 - If it is needed to be checked only for '--with-uuid=bsd',
> should just NetBSD be checked?

I don't see any reason not to check in the BSD code path --- it's
a cheap enough test. On the other hand, in the other code paths
there is no evidence that it's necessary, and we'd find out soon
enough if it becomes necessary.

> 2 - Should it error out without including current UUID version in the
> error message? General error message could mask if the 'uuid_create'
> function starts to produce UUIDs other than version-4.

Yeah, I thought reporting the actual version digit was worth doing,
and made it do so.

Pushed with those changes and doc updates. I did not push the
variant expected-file. I think the entire point here is that
we are *not* deeming the new NetBSD implementation acceptable,
so allowing it to pass regression tests is the wrong thing.

regards, tom lane