Re: [new] multimedia/opustags

On 2020/11/12 10:56, Omar Polo wrote:
>
> Omar Polo <op@omarpolo.com> writes:
>
> > Ashton Fagg <ashton@fagg.id.au> writes:
> >
> >> Omar Polo <op@omarpolo.com> writes:
> >
> > [...snip...]
> >
> >> You can also just use GH_TAGNAME directly instead of V. Just set:
> >>
> >> GH_TAGNAME = 1.5.0
> >>
> >> and I think not defining distname (default) should be OK.
> >
> > You're right, it doesn't need the -${V} on the distname.
>
> ...and I was wrong. PKGNAME defaults to DISTNAME, so, unless we want to
> provide an explicit PKGNAME with the version, DISTNAME should have
> one. (but fetching works even without the explicit version on the
> DISTNAME)

A reasonable default is provided based on GH_TAGNAME in many
cases. All that is needed here is

GH_ACCOUNT = fmang
GH_PROJECT = opustags
GH_TAGNAME = 1.5.0

(no V/DISTNAME). Check "make show=PKGNAMES".

With the various changes your distinfo got out of sync
(opustags.tar.gz with no version), that needs a regen.

(Distfile names are a common problem I've noticed with various
submissions on ports@, the filename often ends up as just the
version number .tar.gz which is not legit, this sometimes comes
from trying to handroll the equivalent of what GH_* already does
- in general it's a good idea to sanity-check the distinfo file
before sending out a diff/tar).

> >> Don't take my word as law though, I'm a newbie myself.
> >>
> >> Hope that helps.
> >>
> >> Ash
> >
> > I'll enable the tests and update port tomorrow.
> > Thank you for testing :)
>
> Here's another try, now with 100% more testing!
>
> - added TEST_DEPENDS and enabled testing
> - moved to audio instead of multimedia
>
> However, some tests fails. In one case the regression suite misuse
> getopt_long (by providing an argv that's not NULL-terminated) and that
> causes the failure.
>
> I've patched out a bit in t/opustags.t: it checks if the locale is set
> to UTF8 expecting an output different from our locale(1). Still, it
> fails 8/47 tests in the final suite (with 4 being skipped), but at least
> one of them fails because it tries to open $EDITOR (and setting
> TEST_IS_INTERACTIVE doesn't help).
>

Setting a UTF-8 environment allows a few more to pass:

TEST_ENV = LC_CTYPE=en_US.UTF-8