ports/36112: [PATCH] New feature for whole ports tree: GS_PORT variable

[Jens Rehsack]

On 28.07.2003 13:43, Kris Kennaway wrote:

> On Mon, Jul 28, 2003 at 12:12:22AM +0200, Tilman Linneweh wrote:
> 
>> Just picking a random PR.
>> Instead of adding knobs to every port, a more
>> generic solution might be appropriate, e.g. a bsd.port.mk patch.
>> 
>> 
>> Oh, and I am not the first one with this idea.
>> 
>> PR 36112 by lev tries to introduce a IMHO better solution.
> 
> I thought I had replied recently to that PR (I'll add this mail as a
> reply to that PR so we can record the rest of this thread).  I also
> like this approach, but would prefer a more verbose variable name
> (i.e. GHOSTSCRIPT_PORT) so that it's less cryptic.
> 
> I'm not sure about the need for GS_PORTSUFFIX though: if someone wants
> to avoid X11 support then they would set GHOSTSCRIPT_PORT to
> print/ghostscript-gnu-nox11 (and set WITHOUT_X11) so that none of
> their ports that use gs pull in the X dependency.  ImageMagick just
> happens to be about the only port that provides for this facility at
> the moment, by checking WITHOUT_X11 explicitly.

So what's your preferred next step: committing patch from ports/36112
and close all PR's I submitted and re-check after commit all ports
which are not affected, or shall I re-submit with the style of
ports/36112?

And to give a statement about WITHOUT_X11 and print/ghostscript-*-nox11:
I think it makes much sense to check, because the right dependency is
kept, but I don't want to blow the check to much and avoid changing
the configure-scripts more than absolutely necessary. The scripts must
have knowledge about the handling of bsd.port.mk and the style how
print/ghostscript-*-nox11 are working. I don't like those kind of code,
so I tried to have a small step and discuss about further changes
with experts (eg. like you and Tilman).

I've seen some problems with ghostscript-fonts (because I don't
understand why it wont work with ghostscript-afpl) and
print/cups-pstoraster (were I've got it depend on
print/ghostscript-afpl but wasn't able to check whether it's
working right or not).

I like the way like lang/php4 handles the right port-dependency,
but I strongly agree to Joe Markus Clarke who wrote to ports/52674
that such an include should be dependent port related and not
be for portmgr in Mk/. I would like to see a bsd.ghostscript.mk,
but not in Mk/ but in sth. like print/ghostscript.

I don't know whether it makes sense to know about such includes
in bsd.*.mk in Mk/ or not, knowing about and including
print/ghostscript/bsd.ghostscript.mk could make porters life much
easier, but as seen with lang/php4 and security/openssl, it works
otherwise, too.

Best,
Jens