svn commit: r484696 - in head: audio/fdmf audio/pulseaudio audio/rsynth audio/zinf databases/dbtool databases/fpc-gdbm databases/gdbm databases/p5-GDBM databases/py-gdbm databases/pypy-gdbm databas...

[Alexey Dokuchaev]

On Wed, Nov 14, 2018 at 12:20:46AM +0100, Mathieu Arnold wrote:
> On Sun, Nov 11, 2018 at 03:38:51PM +0000, Alexey Dokuchaev wrote:
> > New Revision: 484696
> > URL: https://svnweb.freebsd.org/changeset/ports/484696
> > 
> > Log:
> >   Update `databases/gdbm' to version 1.18.1, an important bugfix release
> >   which restores compatibility with old databases (version 1.8) and some
> >   later versions which were built without mmap(2) support.  Due to shlib
> >   version change, bump port revisions of the consumer ports.
> 
> This probably broke lang/perl5.* when GDBM is enabled:

(Since you didn't reply on IRC) This is a known issue since June, quoting
https://rt.perl.org/Public/Bug/Display.html?id=133295:

On 2018-06-26, Petr Pisar <ppisar at redhat.com> wrote:
> So the reason for the test is to check gdbm fatal errors are properly
> passed to Perl code. Provided the only place gdbm uses these fatal errors
> is when storing data and these are hidden by the mmaped I/O since
> gdbm-1.15, I would vote for removing the test as we don't have any
> reliable way to provoke the fatal error.

I suggest the same (remove the test).  I can prepare a patch if you want.

./danfe