[Halld-offline] BMS osrelease stings

[David Lawrence]

Hi Igor,

   I just committed a change to implement your suggestion. At one point 
(more than 5 years ago) it seems there was a reason for using "-p", but 
I don't recall what it was and so don't know if it is still valid.

   I'm CC-ing the offline list on this so if someone does run across 
some bad behavior resulting from this change on their platform, they can 
know the cause and we can address it.

Regards,
-David

On 3/18/12 2:58 AM, Igor Senderovich wrote:
> Hi David,
>
> Our cluster at UConn has a bunch of old Athlon machines, which lead to 
> a problem with the string you get from BMS/osrelease.pl string. An 
> athlon machine admits to being "athlon" when you do `uname -p` but the 
> architecture is still i686 and we don't bother making a separate build 
> on an Athlon machine. So far I have been working around this.
>
> But, for the purposes of just having compatible builds, would it make 
> sense to switch the script to use `uname -m` to quote the architecture 
> (i686, x86_64...) rather than processor? Or would this break something.
> (This was my workaround and I haven't seen problems on the range of 
> machines I've had access to.)
>
> Thanks,
> -Igor