On 2021/08/16 16:59, Theo Buehler wrote:
> rsadowski, I think the X11BASE patch is worth considering.
>
> > I see that upstream now provides release tarballs which is much better.
> > So it's good to ditch the autoconf stuff and use MASTER_SITES.
>
> Probably some confusion in terminology, but what does autoconf have to
> do with release tarballs?
Release tarballs normally have autoconf files generated (so use
CONFIGURE_STYLE=gnu rather than CONFIGURE_STYLE=autoconf or autoreconf
unless the input files need patching), whereas "tarball autogenerated from
git" usually don't. (Except sometimes they do!)
> > +GH_ACCOUNT = resurrecting-open-source-projects
> > +GH_PROJECT = scrot
> > +GH_TAGNAME = 1.6
>
> From Makefile.template:
>
> # /releases/ -> preferred. ignore GH_*, just use MASTER_SITES and DISTNAME.
Yes. Don't set GH_*, don't set DISTFILES, do set MASTER_SITES and DISTNAME.
If GH_* are set, MASTER_SITES should not be set. (Sometimes MASTER_SITES0 etc
might be set if multiple files are needed, but not in this case).
> > ++ if(opt.exec) {
> > ++ if (pledge("cpath exec prot_exec rpath stdio wpath proc", NULL) ==
> > -1) {
>
> This update should remain decoupled from adding pledge, IMO.
Agreed.
> I'm generally skeptical of ports pledges that add prot_exec (except for
> the monsters where it's unavoidable). Can this be reduced to something
> sane by hoisting whatever needs it? Also, please sort the promises to
> canonical order (roughly the order in the manual, look at what is done
> in base).
Also agreed (and I'd really like to see an analysis of functions in the
code and in libraries called by the code to explain why the pledge is
correct... remember the pledge here doesn't just deal with scrot code,
it also deals with code in the libraries, e.g. c Imlib2 X11 Xcomposite
Xext Xfixes).
> > -@@ -148,11 +149,11 @@ name_thumbnail(char *name)
> > + new_title = malloc(length);
> > ++ memset(new_title, 2, length);
>
> This doesn't look right at all.
I think more bugs have been introduced than fixed by adding string handling
patches to ports..
