cvs commit: ports/news/inn Makefile

[Shaun Amott]

On Fri, Aug 25, 2006 at 04:19:18PM -0400, Kris Kennaway wrote:
> On Wed, Aug 23, 2006 at 08:57:28PM +0000, Shaun Amott wrote:
> > shaun       2006-08-23 20:57:28 UTC
> > 
> >   FreeBSD ports repository
> > 
> >   Modified files:
> >     news/inn             Makefile 
> >   Log:
> >   Amend hostname check to succeed if /bin/hostname fails.
> >   Another work-around for those pesky amd64 machines.
> 
> This isn't going to work as you intend since the IGNORE check is run
> on pointyhat to decide whether to build the ports, which naturally
> succeeds.

Well, maybe this wasn't the right approach, but I did intend the check
to affect pointyhat builds.

> The real issue here is not 'amd64 machines' but ports built in a jail
> without access to a DNS server (this will become the default in the
> near future).

I know - I wasn't suggesting it was a problem with amd64 as a platform -
it's just a coincidence that the affected machines have all been amd64.
The port checks that the machine/jail's hostname is valid (i.e.
host.domain), not just resolvable, and the work-around seems to work
in both cases I have seen.

> Why does this port need to look up its hostname during build?  If it
> is hard-coding the result in the package, then it is clearly wrong.
> On the other hand if it not hard-coding the result then it begs the
> question of why do it at all.

The look-up is performed as part of a syntax check on the default config
file. The hostname is inserted into the file, which, as you you say, is
wrong.

I have come up with an alternative solution to this issue, which I shall
apply to the inn ports shortly.

Shaun

-- 
Shaun Amott [ PGP: 0x6B387A9A ]
    Scientia Est Potentia.