Some positive feedback on the responsiveness of bzr(explorer)

[Alexander Belchenko]

10.05.2011 1:06, Martitza Mendez пишет:
> 
> Forgive me, I do not have my previous data in front of me; it is in
> the mailing list archives.
> 
> Either I have become much more patient (unlikely), more forgiving
> (also not very probable) or bzr has become noticeably faster and less
> of a memory hog in the 2.4 series.
> 
> Using my favorite real-life test repository, with 15k files in 700
> folders adding up to 1.5GB, I see the following in bzr explorer:

Thank you for testing!

Your numbers still very big IMO, but to improve it further I need to
see where we spend the time and what parts we can speed-up without
rewriting everything from scratch. I'm going to add special log file
for explorer separate from .bzr.log to be able write there many
internal info about time of execution, as I did it in ad-hoc way when
debugging problem with submit pane. So users like you can collect some
important data and then send it for analysis.

> 
> add -- 9s to find files and put then in the list pane, plus another 58
> seconds to actually add
> 
> refresh view -- 8 seconds
> 
> commit -- 6 sec to load the list plane, plus 5.25 minutes to actually
> commit
> 
> refresh view -- 55 seconds
> 
> This was with the bzr installed as part of the windows standalone
> installer 2.4b1 but overloaded with brand new bzr-explorer rev 510 and
> qbzr rev 1385.
> 
> Maximum memory consumption during commit was 315MB.  Total page faults
> about 800,000 (on a machine with 2GB RAM).

What is "Total page faults"?

> 
> There is still work to be done as Alexander has said, but overall this
> is much better than the bzr 2.2 days.  Thank you and congratulations!
> 
> ~M
>