Re: Comments on policy modifications

[Santiago Vila <sanvila@unex.es>]

On Sun, 13 Sep 1998, Zed Pobre wrote:

> <arch>-<os>:
> 
>     There have been three suggestions on this so far.  Add "hurd" to the
> list of acceptable <os> strings (currently the only acceptable one is
> "linux", add "gnu" to the list of acceptable <os> strings, and for each
> current arch, add <arch.hurd>.  
>     My preference is to go with allowing <os> to be either "hurd" or
> "linux".  Allowing <os> to be "gnu" is confusing (and inaccurate), and
> the <arch.hurd> thing really looks like an ugly hack, and calling it
> part of Linux isn't particularly accurate, either.   Would any serious
> problems arise from going with the "i386-hurd" model?  

Yes: configure scripts may fail horribly.

Please, read my previous message about this. These "Architecture
specification strings" are the input for configure scripts, and we can't
change them gratuitously. This is not a matter of preferences.

-- 
 "ee21ffd6ed27f25ca72ff5771576eac6" (a truly random sig)