On 2021/02/18 21:02, Daniel Dickman wrote:
> On Thu, Feb 18, 2021 at 8:34 PM Theo Buehler <t...@theobuehler.org> wrote:
> >
> > On Thu, Feb 18, 2021 at 08:03:15PM -0500, Daniel Dickman wrote:
> > >
> > >
> > > > On Feb 18, 2021, at 6:19 PM, Theo Buehler <t...@theobuehler.org> wrote:
> > > >
> > > > On Fri, Feb 19, 2021 at 12:09:18AM +0100, Theo Buehler wrote:
> > > >> The project has moved to github and is somewhat active there.
> > > >>
> > > >> In particular it merged a gcc-10 fix. It doesn't do releases, so I just
> > > >> picked HEAD. fomalib.h has modified structs and function signatures,
> > > >> so I bumped the major.
> > > >>
> > > >> I added a BDEP on devel/bison which might be avoidable but I haven't
> > > >> tried.
> > > >>
> > > >> Lightly tested on amd64.
> > > >
> > > > I should just have set DISTNAME.
> > >
> > > I think it would be better to set the package version to 0.9.18 like in 
> > > the project’s Makefile.
> > >
> > > That way if the project ever does do a release we don’t have to bump 
> > > EPOCH.
> > >
> > > Until then we can continue to resync with git head by bumping REVISION.
> >
> > I can do that if you feel strongly about it. I think the date is a more
> > useful indicator than a release from ~6 years back.
> 
> then just add the date at the end to the package version.

e.g. 0.9.18pre20200928

> > What's the problem
> > with bumping EPOCH?
> 
> probably nothing in a leaf port, although in general I try to stay
> away from it given interactions with pkgspec meaning one has to
> remember to bump revisions of consumers when epoch changes.

yes, this. also having this is a common practice (even on leaf ports)
it's more likely people will use it as an example in other ports.

Reply via email to