Antti Harri:

> Makes sense. Can someone check the attached tarball? I'm sure it has room for 
> improvement as I did it pretty hastily. I'm going to call it a day so I'm 
> submitting it now. :-)
> 
> Can some of the pkgpath/conflict markers be combined and are they correct?

They look fine to me.  I think we'll also need quirk entries, won't
we?

These ports will need to have their dependencies updated:
  editors/libreoffice
  graphics/makehuman
  sysutils/backintime/common

> Does the .so in libexec need special handling?

Theoretically, this and the rest of the (g)stdbuf command should
go into PFRAG.shared.  In practice, I can't find any provisions for
platforms without dynamic linking.  We could disable the port for
NO_SHARED_ARCHS, but I think it's better to allow building it there
and accept that some parts won't work.  The sole affected arch,
vax, can't uncompress the tarball anyway.

Some further comments:

I think the port should go into sysutils.  That's probably sufficient
for CATEGORIES, too.

Why did you add NO_REGRESS?  This is wrong.

The configure script tries to pick up gmp.  I have no particular
opinion whether we want this, it only affects the factor and expr
commands, but we should either fully enable it (LIB_DEPENDS,
CPPFLAGS/LDFLAGS) or disable it (--without-gmp).

-- 
Christian "naddy" Weisgerber                          [email protected]

Reply via email to