darcs

Issue 1889 darcs get --lazy is slower with packs than without

Title darcs get --lazy is slower with packs than without
Priority bug Status resolved
Milestone 2.8.0 Resolved in
Superseder Nosy List dmitry.kurochkin, exlevan, kowey, mornfall
Assigned To kowey
Topics HTTP, Performance, Regression

Created on 2010-07-12.15:01:06 by kowey, last changed 2010-08-04.22:16:09 by kowey.

Messages
msg11725 (view) Author: kowey Date: 2010-07-12.15:01:03
I'm just creating this ticket so that we don't accidentally release Darcs 
2.5 with a packs feature that makes --lazy slower.

In the worst case, we can just hide the packs feature (or mark it as 
experimental).
msg11955 (view) Author: mornfall Date: 2010-08-04.20:49:48
Eric, could you please check again with the current HEAD? I have pushed a 
patchset from Alexey which should help with this a lot. Thanks!
msg11962 (view) Author: kowey Date: 2010-08-04.22:16:08
On Wed, Aug 04, 2010 at 20:49:49 +0000, Petr Ročkai wrote:
> Eric, could you please check again with the current HEAD? I have pushed a 
> patchset from Alexey which should help with this a lot. Thanks!

With pleasure.  After optimize --http with the new code to get the
smaller packs tarball:

darcs get http://darcs.net --lazy --no-cache

On my Mac on home cable modem

  darcs-2.4.98.2:    real  105.99   user 1.15   sys 1.48 
  HEAD          :    real   17.37   user 0.94   sys 0.71  

On my Linux machine at University:

  darcs-2.4.98.2     real  134.60   user 0.66   sys 0.40
  HEAD               real   11.78   user 0.95   sys 0.32

Congratulations, Alexey!

PS: I think this is blog worthy... full timings may be good for that
too.

-- 
Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow>
For a faster response, please try +44 (0)1273 64 2905.
History
Date User Action Args
2010-07-12 15:01:06koweycreate
2010-07-25 14:29:59tux_rockersetmilestone: 2.5.0 -> 2.8.0
2010-08-04 20:49:49mornfallsetassignedto: exlevan -> kowey
messages: + msg11955
nosy: + mornfall
2010-08-04 22:16:09koweysetstatus: needs-reproduction -> resolved
messages: + msg11962