Binary differencing performs poorly (erractically) on very large text file (REPOST)

I have a large text file (around 47 MB) which is a database dump
(created by msqldump). I periodically commit it to an SVN repo.
Sometimes the binary differencing works just fine and I get a small
sized revision in the repo. Other times I get a "full" sized revision
in the repo, that is revision that is compressed, but essentially the
same size I get when committing the file to a virgin repo.

Doing a "diff" on the client side files always generates a "relatively"
small set of differences.

Observations:

1) Something about the binary differencing is working erratically,
perhaps

   only for this large file, in the sense of not always noticing that
the

   differences between two revs are much smaller than the entire file.

2) Compression, on the other hand, seems fine.

Some numbers:

Repo Client Client Diff

Rev (KB) File (KB) Diff (KB)

-------- --------- ---------

11,532 53,942 n/a

11,532 53,943 166

   335 53,961 656

11,594 54,023 244

    89 54,023 1

Reproducible on a variety of clients and servers, most recent of which
is 1.4.2 running on RHEL/CentOS 5.1

Questions:

1) Is this a known problem?

2) Is there some easy way I could force the client and/or server to spew
out

   debug information that would be of assistance to some guru?

3) Should I open a problem report, or just send this to the "dev" list?

Thanks

-- Charlie Butterfield

 
Received on 2008-02-16 00:49:36 CET