[MPlayer-users] Configure severely broken on IRIX

The Wanderer inverseparadox at comcast.net
Wed May 6 22:45:07 CEST 2009


On 05/06/2009 04:13 PM, Matthew Saunier wrote:

> Hi everyone, mplayer doesn't even try to compile on IRIX. I haven't
> dug into the code yet, but I have fixed your configure. It is not, as
> the slashbang indicates, a legal sh script. Under bash it mostly
> works, but it assumes GCC for everything and is incapable of
> detecting the architecture of SGI machines. I've added MIPSpro
> compiler support to it and cleaned up some bad assumptions and
> linux-isms, patch attached.

Patches should be sent to mplayer-dev-eng, following the rules in
DOCS/tech/patches.txt.

Also, MPlayer's configure script does not (or should not) require bash;
it does, however, require POSIX. Any POSIX-compliant /bin/sh should be
able to run it just fine.

> @@ -1327,7 +1327,7 @@
>  # Determine our OS name and CPU architecture
>  if test -z "$_target" ; then
>    # OS name
> -  system_name=$(uname -s 2>&1)
> +  system_name=`uname -s 2>&1`

The specification at

http://www.opengroup.org/onlinepubs/9699919799/utilities/sh.html

(which I understand to be part of POSIX; correct me if I'm wrong) uses
the $(subshell) syntax in at least one of its examples, and does not use
the `subshell` syntax at all. This change, and those like it, are
unlikely to be accepted; indeed I'm fairly sure I've seen commits which
changed it in the other direction.

configure is also not supposed to assume GCC - I remember reading that
people have compiled successfully with e.g. ICC, at the very least. I'm
somewhat skeptical that the sole compiler-related change which I see in
your patch will be accepted as is, but there's pretty good odds that
something with the same goal may go in if you propose it.

-- 
       The Wanderer

Warning: Simply because I argue an issue does not mean I agree with any
side of it.

Secrecy is the beginning of tyranny.


More information about the MPlayer-users mailing list