Hello Ian,

On the machine that runs the tnaur PPC OSX 2 builders, I recently switched 
over to darcs-2.0.0 to be able to do darcs buildbotting. Since then, the time 
spent by the ghc builders in darcs get operations seems to have grown 
significantly. Here are some 
recent measurements for STABLE

2008/05/26-02:51 - 2008/05/26-04:49: elapse 118 
['/Users/thorkilnaur/tn/install/darcs-2.0.0/bin/darcs','get','--verbose','-
-partial','--repo-name','build','http://darcs.haskell.org/ghc-stable/ghc/'] 
signal None, exit 0
2008/05/28-02:51 - 2008/05/28-05:11: elapse 140 
['/Users/thorkilnaur/tn/install/darcs-2.0.0/bin/darcs','get','--verbose','-
-partial','--repo-name','build','http://darcs.haskell.org/ghc-stable/ghc/'] 
signal None, exit 0
2008/05/30-02:51 - 2008/05/30-05:19: elapse 148 
['/Users/thorkilnaur/tn/install/darcs-2.0.0/bin/darcs','get','--verbose','-
-partial','--repo-name','build','http://darcs.haskell.org/ghc-stable/ghc/'] 
signal None, exit 0

and for HEAD:

2008/05/27-02:50 - 2008/05/27-03:23: elapse 33 
['/Users/thorkilnaur/tn/install/darcs-2.0.0/bin/darcs','get','--verbose','--
partial','--repo-name','build','http://darcs.haskell.org//ghc/'] signal None, 
exit 0
2008/05/29-02:50 - 2008/05/29-05:53: elapse 183 
['/Users/thorkilnaur/tn/install/darcs-2.0.0/bin/darcs','get','--verbose','-
-partial','--repo-name','build','http://darcs.haskell.org//ghc/'] signal None, 
exit 0
2008/05/31-02:52 - 2008/05/31-05:02: elapse 130 
['/Users/thorkilnaur/tn/install/darcs-2.0.0/bin/darcs','get','--verbose','-
-partial','--repo-name','build','http://darcs.haskell.org//ghc/'] signal None, 
exit 0

The tnaur x86 Linux builders still use darcs-1.0.9 (mainly because I haven't 
restarted the session that runs the buildbot slave since switching over to 
darcs-2.0.0) and here the times are significantly smaller (STABLE and HEAD 
measurements mixed):

2008/05/28-03:28 - 2008/05/28-03:36: elapse 8 
['/home/tn/tn/install/darcs-1.0.9/bin/darcs','get','--verbose','--partial','-
-repo-name','build','http://darcs.haskell.org/ghc-stable/ghc/'] signal None, 
exit 0
2008/05/29-03:28 - 2008/05/29-03:43: elapse 15 
['/home/tn/tn/install/darcs-1.0.9/bin/darcs','get','--verbose','--partial','
--repo-name','build','http://darcs.haskell.org//ghc/'] signal None, exit 0
2008/05/30-03:28 - 2008/05/30-03:39: elapse 11 
['/home/tn/tn/install/darcs-1.0.9/bin/darcs','get','--verbose','--partial','
--repo-name','build','http://darcs.haskell.org/ghc-stable/ghc/'] signal None, 
exit 0
2008/05/31-03:28 - 2008/05/31-03:35: elapse 7 
['/home/tn/tn/install/darcs-1.0.9/bin/darcs','get','--verbose','--partial','-
-repo-name','build','http://darcs.haskell.org//ghc/'] signal None, exit 0

Comparing 
http://darcs.haskell.org/buildbot/all/builders/tnaur%20PPC%20OSX%20head%202/builds/94/steps/darcs/logs/stdio
and 
http://darcs.haskell.org/buildbot/all/builders/tnaur%20x86%20Linux%20head/builds/194/steps/darcs/logs/stdio
 
the darcs-2.0.0 get seems to be dealing with about 18,500 patches, whereas 
the darcs-1.0.9 get deals with only 15 (the HEAD was tagged recently, I seem 
to recall).

This is not really a problem, but it seems that it could point to some darcs 2 
problem or a problem with the way the ghc buildbot uses darcs. So I would be 
very interested in hearing your comments to this.

Thanks and best regards
Thorkil

_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc

Reply via email to