ports/87853: [fix] multimedia/mplayer: no bsdbt848 driver compiled in

[Thomas E. Zander]

The following reply was made to PR ports/87853; it has been noted by GNATS.

From: "Thomas E. Zander" <riggs at rrr.de>
To: Simun Mikecin <numisemis at yahoo.com>
Cc: bug-followup at FreeBSD.org, Edwin Groothuis <edwin at FreeBSD.org>
Subject: Re: ports/87853: [fix] multimedia/mplayer: no bsdbt848 driver compiled in
Date: Fri, 4 Nov 2005 09:50:42 +0100

 --UPT3ojh+0CqEDtpF
 Content-Type: text/plain; charset=iso-8859-15
 Content-Disposition: inline
 
 Hi,
 
 > So, my conclusion is: on the architectures that have
 > their "machine" version of ioctl_bt848.h (and those
 > are alpha, i386 and pc98), those "machine" version
 > files get installed as
 > /usr/include/machine/ioctl_bt848.h.
 > But *ALL* architectures get
 > /usr/include/dev/bktr/ioctl_bt848.h installed.
 
 omg...I see where this (sorry) crap is leading us to.
 
 Don't worry, I understand your problem and this is certainly needed to
 be handled. I also didn't intend to refuse any change to the port, it
 is just that I can't approve to this particular pr which only means
 that we have to modify it.
 
 Despite the fact that it is really a horrible situation to have
 a different include base for these two (very similar!) architectures
 in the same release, there is another problem which forbids to just
 replace the location of ioctl_bt848.h
 We have to address this depending on ${OSVERSION} because RELENG_4
 still should be supported by the ports which allow us to do so and this
 branch iirc only knows /usr/include/machine/ioctl_bt848.h!
 
 So I suggest that we proceed as follows:
 - drop the xmms and cdrom/dvdrom patches your pr is covering as we do
   this already in post-patch:
 - don't replace the location of this header in the configure script,
   instead extend it by something like:
   ...
   #elif (__FreeBSD__ >= 5)
   #include <dev/bktr/ioctl_bt848.h>
   ...
 
 If you're okay with that, I'll prepare a diff and see that it gets
 committed.
 
 Regards,
 Riggs
 
 P.S. For the record: Finally it gets a mess to maintain ports like this
 one if there is nothing you can really rely on. I don't have enough
 boxes to genuinely check everything on every architecture for every
 RELENG_*. I don't have a RELENG_4 box, no amd64, no ia64, no sparc and
 so on. Still a port maintainer should support every officially supported
 branch (RELENG_4-7 !!) on every machine and as we see from time to
 time, they *do* differ which makes it completely impossible for a
 maintainer to reproduce all the problem a user might have with a specific
 version or branch. This is definitely *not* an issue to be handled
 sloppily. Edwin, maybe you could discuss this problem with some src
 committers some time before things get worse...
 
 --UPT3ojh+0CqEDtpF
 Content-Type: application/pgp-signature
 Content-Disposition: inline
 
 -----BEGIN PGP SIGNATURE-----
 
 iD8DBQFDayDijdSJKchZls0RAun0AJ9CailDPWWm0av2ALZgkpRd/IMKqgCfTTCq
 02ZmrrHL0ed7MU6Ti6fCGSc=
 =b/nE
 -----END PGP SIGNATURE-----
 
 --UPT3ojh+0CqEDtpF--