ps, this is a nice example of the sort of thing that is good to get into
the tree around this time. the port is already broken so it can't be
made much worse, there's a simple fix with few downsides, that missed
getting committed before.
much better to spend time around now on things like this than pushing
in new or more complicated/risky changes.
On 2024/09/11 00:39, Stuart Henderson wrote:
> On 2024/09/10 16:12, A Tammy wrote:
> >
> > On 9/10/24 1:45 PM, Rubén Llorente wrote:
> > > Hi there,
> > >
> > > it would be a good idea to have this fix included in lrzip before the
> > > next -release.
> > >
> > > ###
> > > Hello,
> > >
> > > lrzip's ZPAQ has been broken since this port got added to the tree.
> > >
> > > This patch fixes it. It has been working for years for me.
>
> Diff was mangled so I retyped it by hand
>
> > > diff -u -p -r1.3 Makefile
> > > --- Makefile 4 Nov 2022 15:05:10 -0000 1.3
> > > +++ Makefile 22 Apr 2024 11:12:34 -0000
> > > @@ -3,7 +3,7 @@ COMMENT = compression utility for large
> > > GH_ACCOUNT = ckolivas
> > > GH_PROJECT = lrzip
> > > GH_TAGNAME = v0.651
> > > -REVISION = 0
> > > +REVISION = 1
> > >
> > > CATEGORIES = archivers
> > >
> > > @@ -24,7 +24,7 @@ RUN_DEPENDS = shells/bash
> > >
> > > CONFIGURE_STYLE = autoreconf
> > > CONFIGURE_ARGS = --disable-doc
> > > -CONFIGURE_ENV = CPPFLAGS="-I${LOCALBASE}/include" \
> > > +CONFIGURE_ENV = CPPFLAGS="-I${LOCALBASE}/include -DNOJIT" \
> > > LDFLAGS="-L${LOCALBASE}/lib" \
> > > ac_cv_prog_ASM_PROG='no '
> > > # don't pick up archivers/nasm; it breaks build
> > >
> > > ##
> > >
> >
> > How do we make sure this hasn't broken anything else?
> >
> > I'm not seeing tests running on this port. Can you try to get the tests
> > to pass?
>
> The test script (regressiontest.sh) is full of gnuisms. If you place
> symlinks wc -> gwc, tar -> gtar, sort -> gsort, head -> ghead in the
> path ahead of /usr/bin and run it under bash you'll get some of it
> to work, but there are some other differences around readonly dirs
> etc that will cause some failures.
>
> -DNOJIT is definitely better than not so I will go ahead and commit.
