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

Re: Frequent revisions to large compressed files

From: Ryan Schmidt <subversion-2008c_at_ryandesign.com>
Date: Tue, 2 Dec 2008 14:21:45 -0600

On Dec 2, 2008, at 13:09, Kynn Jones wrote:

> Hi. I would like to put some large compressed files under version
> control using Subversion. The largest of these files is about
> 200MB and the smallest is about 40MB, and they are all compressed
> using gzip, and I expect that I will need to commit new revisions
> of these files at the rate of once a day.
>
> Does Subversion keep track of changes incrementally? And, if so,
> would small incremental changes in the *uncompressed* versions of
> these files translate to small incremental changes in the depository?
>
> I suspect that the answer to the last question is "no", because (I
> guess) that, for it to be "yes", Subversion would have to be smart
> enough to realize that it is dealing with a compressed file and
> treat the contents accordingly. This seems like too much to expect...
>
> My concern here is that as new revisions of these files are
> committed to the depository, its disk footprint will balloon
> rapidly, because, even though the uncompressed revisions are minor,
> the compressed file will change drastically from one revision to
> the next, so that even an incremental update will require having
> almost the entire file saved each time.
>
> (I hope the above made sense!)

You're correct in all you say. It in fact depends on the compression
algorithm; if it causes the old and new files to look totally
different, then Subversion has no choice but to store all of that
difference. Subversion does not examine your binary files or attempt
to decompress them.

> Any words of wisdom on frequently committing large compressed files
> to Subversion would be much appreciated!

Either store the uncompressed files in the repository (yes the
initial revision will be larger but subsequent revisions will be only
as big as the changes so this should be better over time than regular
compressed files), or compress the files in a way that causes only
the differences in the original files to show through in the
compressed files. For example, for gzip, you can use the --rsyncable
flag. gzip requires a patch to implement this flag; you can get it here:

http://trac.macports.org/ticket/16446

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=912494

To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_subversion.tigris.org].
Received on 2008-12-02 21:22:45 CET

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.