[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Multi-path updates

From: Erik Huelsmann <ehuels_at_gmail.com>
Date: 2007-10-23 22:42:42 CEST

> I just ran an interesting comparison of 2 raw svn update commands. Both
> update the nearly same set of data but the runtimes are drastically
> different.

>
> % time svn up
>
> 0.160u 0.320s 0:27.81 1.7% 0+0k 0+0io 1366pf+0w
>
>
>
> % time svn up -N * */*
>
> 31.440u 10.440s 14:38.91 4.7% 0+0k 0+0io 873409pf+0w

The latter is not the same as 'svn up', but rather

$ svn up -N file-1
...
...
$ svn up -N dir-N/file-M

Which sleeps for 1 second after every 'svn up' call. It's called
"sleeping for timestamps". We do it so that we know that the clock has
advanced enough for any script modifying the working copy will
actually advance the last-modified time on the modified file (on Unix,
many filesystems have a 1s timestamp resolution).

> The second run is a 32x increase in run time. Now my particular issue is
> not quite the same as this (more to do with working around the lack of
> –depth in the 1.4.x clients) but this seems seriously out of whack.

I hope the above explains it...

bye,

Erik.
Received on Tue Oct 23 22:43:13 2007

This is an archived mail posted to the Subversion Users mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.