Re: +CONTENTS issue with lang/perl5 update

[Chavdar Ivanov <ci4ic4%gmail.com@localhost>]

On Tue, 12 Dec 2023 at 00:37, Chavdar Ivanov <ci4ic4%gmail.com@localhost> wrote:
>
> On Mon, 11 Dec 2023 at 22:55, Greg Troxel <gdt%lexort.com@localhost> wrote:
> >
> > Chavdar Ivanov <ci4ic4%gmail.com@localhost> writes:
> >
> > > To chip in, my present pkg_rr fails as well in several packages
> > > showing similar gssapi errorw, namely samba4:
> >
> > As always, I ask that people separate "pkg_rr problems" from "make
> > replace" problems.  All pkg_rr does is run make replace in what is
> > believed to be the correct order.  If the order is correct, and the make
> > replace fails, it's not a pkg_rr problem.  If the order is wrong, or
> > things are mishandled (like multiversion packages), then it is.
> >
> > > [1936/3885] Compiling source3/librpc/crypto/gse.c
> > > In file included from
> > > ../../third_party/heimdal/lib/gssapi/gssapi/gssapi_spnego.h:39,
> > >                  from ../../lib/replace/system/gssapi.h:47,
> > >                  from ../../lib/krb5_wrap/gss_samba.h:27,
> > >                  from ../../auth/kerberos/pac_utils.h:27,
> > >                  from ../../source3/librpc/crypto/gse.c:38:
> > > ../../third_party/heimdal/lib/gssapi/gssapi/gssapi.h:94:40: error:
> > > conflicting types for 'gss_name_t'
> > >    94 | typedef struct gss_name_t_desc_struct *gss_name_t;
> > >       |                                        ^~~~~~~~~~
> > > .....


Weird, but that's what happened with this system. It turned out I had
compat80 and compat90 installed, the links they place in /lib and
/usr/lib were screwing the samba build.

I don't remember why I had installed them perhaps a year ago; nothing
was depending on them.

Just FYI.

> >
> > My guess is that somehow two versions of kerberos are being included.
> > But that's very guessing.
> >
> > Certainly you should clean all workdirs and do "make" in samba4, without
> > involving pkg_rr, and report that.
>
> That's exactly what I did - this was a separate from pkg_rr attempt. I
> don't know if there is any particular problem with pkg_rr in this
> case, although I had some weird results prior to the problem with
> samba4 ( I have lang/ghc 9.4 installed - because of pandoc - and for
> some reason it rebuilt the same version perhaps three times in a row
> after a failure elsewhere; I ended up removing ghc and everything
> depending on it and installing the latest version etc...). Anyway, in
> my case I think I am close to nuking whatever is installed via pkgsrc
> and rebuilding it from scratch, which is not that big deal, it will
> just take time.  I also had a problem with cairo 1.18nb1 - during a
> pkg_rr run. it would not replace the previous 1.16 version as there
> was a single .h file which appeared in another package - this sort of
> version jumps do cause sometimes problems and have to be carried out
> manually. Otherwise, pkgclean, pkg_admin rebuild[-tree] and pkg_chk
> -uq are ran every time I decide to go through another few days of
> pkg_rr.... But then, I usually run -current
>
>
>
> --
> ----



-- 
----