2010/10/6 Lluís Batlle i Rossell <virik...@gmail.com>

> Hello,
>
> we have a fossil repository of less than 100 commits, with a checkout of
> around
> 300MB, and around 3000 files.
>
> The fossil repository size is about 150MB.
>
> We notice that the 'fossil changes' command is very fast, but 'fossil
> commit'
> lasts for a few seconds.
> Can it be that 'fossil changes' checks the mtime, while 'fossil commit'
> reads
> all the files?
>

The default setting for Fossil is to use mtime to figure out which files
have changed.  (You can use "fossil setting mtime-changes off" to cause
Fossil to compute a SHA1 checksum over the content of every file in order to
see which files have changed, if you don't trust mtime.)  The use of mtime
allows "fossil changes" and "fossil status" to be quick.

But on a "fossil commit", Fossil computes an MD5 checksum over every file in
the repository, twice.  It computes an MD5 checksum over all files as they
are contained in the repository, and it computes the MD5 checksum over all
files as they exist on disk, and it compares both results to the R-card in
the manifest.  This is a safety check.  Everything has to be right or the
commit will rollback.

This can take a few seconds.  On the other hand, commits are not that
common.  (A few times per day?)  And Fossil goes by the philosophy that
there is no merit in getting the wrong answer quickly.

More information on the cross-checks that Fossil performance can be seen
here:

    http://www.fossil-scm.org/fossil/doc/tip/www/selfcheck.wiki

These self-checks help to ensure that Fossil never loses or corrupts
content, which is an important property for a VCS.

>
> Does the commit time depend mostly on the checkout size, or the repository
> size?
> If we import all the svn history into our fossil repository, it will mean a
> 200%
> increase in repository size but a 0% increase in checkout size, and we
> doubt
> whether to import it or not.
>
> Thank you,
> Lluís.
>
> _______________________________________________
> fossil-users mailing list
> fossil-users@lists.fossil-scm.org
> http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>



-- 
D. Richard Hipp
d...@sqlite.org
_______________________________________________
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Reply via email to