[developers] More problems with PET

[Christopher Rupp]

Hi Bernd,

kiefer at dfki.de said:
> I just checked in a change that should fix the problem. I'm waiting for  CJ's
> reaction, though.

Sorry, I couldn't react to this in real time. I've done an SVN update and a new
make. What I got initially was an error on a missing 
`mrs-initial-semantics-path'
setting. I put the inclusion of mrs.set back in common.set (I had thought that
was redundant because it's in english.set too) and now I get MRSs again. I hope
this is sufficient positive feedback for now. I will try to test this some more
today but the fix looks right.

Another small point for this audience: the file names set in common.set 
apparently
have to be absolute. What I know mainly is 'smaf-conf'. That's not very helpful
under version control. Is this just one badly coded example or a general 
policy?
Is a variant with some kind of relative path available (to compare with the
include mentioned above). Sorry to mix this up with a practical bug fix, but
I'm making my environment available to a wider user group and it would be 
cleaner
if this was more anonymous.

Cheers,

C.J.

-- 
Dr. C.J. Rupp

University of Cambridge Computer Laboratory
William Gates Building
15 JJ Thomson Avenue
Cambridge CB3 0FD, UK

Tel: +44 1223 767025 
Fax: +44 1223 334678
Mobile: +44 795-8496 916
Email: cr351 at cl.cam.ac.uk