>> I have run a couple of tests -
>> . A script to do 500 commits of random line of text to a file (fsfs):
>> CVS Repository: 764k
>> SVN Repository: 4.3M
>>
>> . A conversion of one cvs "repository" using cvs2svn:
>> (trunk only, fsfs, ~15,000 revisions)
>> CVS Repository: 109M
>> SVN Repository: 614M
>>
>> Why am I seeing such bloated repositories? Svn is using 5-6x the disk
>> space when I expected to see just the opposite.
>>
>> Is this a case where the changesets are small in comparison to entries
>> list?
> I do think this is an artefact of your contrived test.
>
> An fsfs conversion of a 2.6GB CVS repository has resulted in a 2.0GB
> SVN repository of 25,000 revisions.
>
> Assuming commits are typically not single-file one-liners, we expect
> this trend to continue (though our rate of revisions will increase as
> we move to more branching and merging).
I see from more recent posts that space consumption could be due to a
very flat folder structure with a great many files in each folder.
Our repositories are the complete reverse, ~20,000 files in ~5,000
folders (numbers are very approximate, I haven't done a recent check)
- so looking at an average folder, that folder will contain a
typically small number of elements.
It's interesting that this makes such a significant difference, and
incentive to discover a more compact means of storing these
element-lists.
--
Talden
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: users-help_at_subversion.tigris.org
Received on 2008-08-14 00:39:15 CEST