ptlib build failure - breaks pwlib - hence also asterisk - opal - & openh323

[David Southwell]

> On 2010-Sep-23 12:22:51 -0400, Jerry <freebsd-ports.user at seibercom.net> 
wrote:
> >A few months ago, after upgrading to version 8/amd64 and installing
> >OpenSSL from ports, I had several ports bomb out when I attempted to
> >build them.  I filed PR's against them and contacted the maintainers. It
> >appeared that the majority of these port maintainers were not even aware
> >that their port would not build in the presents of OpenSSL when it was
> >installed via the ports tree. In any case, I was able to get them to
> >fix their ports to build correctly.
> 
> So it sounds like everything worked as expected.  I fail to see any
> deficiency in the way things worked.
> 
> >What amazed me is that this is such a common occurrence that it should
> >have been contemplated when the port was released.
> 
> Why do you think that?  It's quite likely that the port maintainers
> didn't install OpenSSL from ports - I know I don't.
> 
> > Perhaps the Porters
> >
> >Handbook should list ports that have a corresponding base system
> >counterpart; thereby, alerting the maintainer that he/she should test
> >against both versions to insure full compatibility.
> 
> I disagree.  I'd be surprised if any part of the base system _didn't_
> have one or more corresponding equivalents in ports.  As well as
> OpenSSL, there are 5 versions of gcc (and probably a half-dozen other
> C compilers), binutils, 5 versions of bind (and several other DNS
> implementations that presumably also implement resolver libraries), 3
> Kerberos variants, 2-4 versions of ncurses, readline, etc.  Expecting
> a ports maintainer to check that their port works with all of these on
> 3 different FreeBSD branches and about 5 different architectures is
> completely unrealistic.
> 
> IMHO, the expectations on a port maintainer are that they verify that
> the port works as expected in their environments (where any unusual
> configurations are listed in the port dependencies) and at least
> builds on all other supported branches/architectures (via tinderbox or
> similar).  Any issues beyond that realistically need to be dealt with
> on a case-by-case basis.
Frankly I have never heard such a load of inconsiderate thoughtlessness and 
denial of responsibility form anyone before.

Installation of ssl from ports is part of a normal configuration. Mainainers 
need to take account of reality.

David

Photographic Artist
Permanent Installations & Design
Creative Imagery and Advanced Digital Techniques
High Dynamic Range Photography & Official Portraiture
Combined darkroom & digital creations
& Systems Adminstrator for the vizion2000.net network