On 2021/11/13 11:48, Klemens Nanni wrote:
> On Sat, Nov 13, 2021 at 11:31:57AM +0000, Stuart Henderson wrote:
> > We have a few ports that use GH_* and have a second distfile, so have to
> > build up their own DISTFILES. Where that uses GH_TAGNAME it's not too bad
> > but the ones with GH_COMMIT are a bit horrible:
> >
> > $ ag -G Makef ^DISTF.*GH_COMMIT
> > devel/cabal-bundler/Makefile
> > 13:DISTFILES =
> > ${DISTNAME}-${GH_COMMIT:C/(........).*/\1/}${EXTRACT_SUFX}{${GH_COMMIT}${EXTRACT_SUFX}}
> > \
> >
> > devel/msbuild/Makefile
> > 29:DISTFILES = ${DISTNAME}{${GH_COMMIT}}${EXTRACT_SUFX} \
> >
> > net/arp-scan/Makefile
> > 22:DISTFILES =
> > ${DISTNAME}-${GH_COMMIT:C/(........).*/\1/}${EXTRACT_SUFX}{${GH_COMMIT}${EXTRACT_SUFX}}
> > \
> >
> > net/ntopng/Makefile
> > 19:DISTFILES= ntopng-${NTOPNG_V}-{}${GH_COMMIT}.tar.gz
> > ndpi-${NDPI_V}-{}${NDPI_COMMIT}.tar.gz:0
> >
> > x11/gnome/gdm/Makefile
> > 15:DISTFILES=
> > ${DISTNAME}-${GH_COMMIT:C/(........).*/\1/}${EXTRACT_SUFX}{${GH_COMMIT}${EXTRACT_SUFX}}
> > \
> >
> > x11/mruby-zest/Makefile
> > 45:DISTFILES= zyn-fusion-{}${GH_COMMIT}${EXTRACT_SUFX} \
> >
> > Seems like it might be useful to provide the generated distfile name
> > in a variable that can be reused here. I set it for ports using either
> > GH_TAGNAME and GH_COMMIT so that it doesn't need modifying if we
> > need to update something to an untagged checkout.
>
> I like the idea!
>
> > Any comments? OK?
> >
> >
> > Index: src/share/man/man5/bsd.port.mk.5
> > ===================================================================
> > RCS file: /cvs/src/share/man/man5/bsd.port.mk.5,v
> > retrieving revision 1.546
> > diff -u -p -r1.546 bsd.port.mk.5
> > --- src/share/man/man5/bsd.port.mk.5 8 Oct 2021 13:52:28 -0000
> > 1.546
> > +++ src/share/man/man5/bsd.port.mk.5 13 Nov 2021 11:27:24 -0000
> > @@ -1925,6 +1925,12 @@ Account name of the GitHub user hosting
> > .It Ev GH_COMMIT
> > SHA1 commit id to fetch.
> > It is an error to specify ${GH_COMMIT} when ${GH_TAGNAME} is specified.
> > +.It Ev GH_DISTFILE
> > +Set by
> > +.Nm
> > +to the generated name of the distribution file.
> > +This can be useful for ports listing multiple
> > +.Ev DISTFILES .
> > .It Ev GH_PROJECT
> > Name of the project on GitHub.
> > .It Ev GH_TAGNAME
>
> OK kn
Thanks.
> > Index: ports/infrastructure/mk/bsd.port.mk
> > ===================================================================
> > RCS file: /cvs/ports/infrastructure/mk/bsd.port.mk,v
> > retrieving revision 1.1558
> > diff -u -p -r1.1558 bsd.port.mk
> > --- ports/infrastructure/mk/bsd.port.mk 8 Nov 2021 13:36:25 -0000
> > 1.1558
> > +++ ports/infrastructure/mk/bsd.port.mk 13 Nov 2021 11:27:24 -0000
> > @@ -612,6 +612,7 @@ GH_PROJECT ?=
> >
> > .if !empty(GH_PROJECT) && !empty(GH_TAGNAME)
> > DISTNAME ?=
> > ${GH_PROJECT}-${GH_TAGNAME:C/^(v|V|ver|[Rr]el|[Rr]elease)[-._]?([0-9])/\2/}
> > +GH_DISTFILE =
> > ${GH_PROJECT}-${GH_TAGNAME:C/^(v|V|ver|[Rr]el|[Rr]elease)[-._]?([0-9])/\2/}${EXTRACT_SUFX}
> > .endif
> >
> > PKGNAME ?= ${DISTNAME}
>
> You're not using GH_DISTFILE for ports using GH_TAGNAME, is that
> intentional?
>
> With your diff, picking a random GH_TAGNAME por:
>
> $ make -C /usr/ports/security/hydra -p | egrep
> '^(DIST(FILES|NAME)|GH_DISTFILE)'
> DISTFILES = ${DISTNAME}${EXTRACT_SUFX}
> DISTNAME =
> ${GH_PROJECT}-${GH_TAGNAME:C/^(v|V|ver|[Rr]el|[Rr]elease)[-._]?([0-9])/\2/}
> GH_DISTFILE =
> ${GH_PROJECT}-${GH_TAGNAME:C/^(v|V|ver|[Rr]el|[Rr]elease)[-._]?([0-9])/\2/}${EXTRACT_SUFX}
I think I see what you mean. GH_TAGNAME doesn't normally set DISTFILES
directly, only DISTNAME (which normal ports infrastructure turns into
DISTFILES by adding EXTRACT_SUFX). So using GH_DISTFILE to build up
that name doesn't work (unless we strip off EXTRACT_SUFX but that
doesn't make sense to me).
Whereas with GH_COMMIT, DISTNAME isn't automatically set at all, it
is up to the port to do so.
Originally I only had this set when you use GH_COMMIT not GH_TAGNAME,
on the basis that it's easy to get the distfile name for a GH_TAGNAME
port anyway. But then I wanted to test an update of librenms from a
commit hash rather than the usual tag, and realised it would reduce
churn in a port Makefile if it could cope with changing between the
two.
So in a nutshell (assuming I understand the question correctly):
yes it's intended.
