repository backup size - cvs vs hg
Andrea Arcangeli
andrea at suse.de
Thu Jun 29 15:26:00 CDT 2006
On Thu, Jun 29, 2006 at 03:53:35PM -0400, Chris Mason wrote:
> Could you please du -ka on the .hg directory? Is most of the space taken up
> by a few big revlogs?
As an example this is the backup of the klive repo:
klive-cvs.tar.bz2 -> 55912 bytes (tar.bz2 of the CVSROOT/klive directory)
klive-hg-bundle -> 127084 bytes (hg bundle ../klive-hg-bundle ../empty
klive-hg.tar.bz2 -> 231691 bytes (tar.bz2 of the klive/.hg dir, no data)
du -ka| sort -nr |head -n 50
476 .
320 ./data
104 ./data/server
72 ./data/xml
72 ./00manifest.i
68 ./00changelog.i
60 ./data/scripts
40 ./data/server/web.py.i
36 ./data/client
16 ./data/server/pgsql.py.i
16 ./data/pgsql
12 ./data/xml/root_page.xml.i
12 ./data/client/klive.tac.i
8 ./data/xml/klive.css.i
8 ./data/server/klive.py.i
8 ./data/pgsql/create.sql.i
8 ./data/COPYING.i
8 ./data/client/klive.sh.i
8 ./data/client/klive.py.i
4 ./undo.dirstate
4 ./undo
4 ./dirstate
4 ./data/xml/vendor.xml.i
4 ./data/xml/rss.xml.i
4 ./data/xml/pci.xml.i
4 ./data/xml/mod.xml.i
4 ./data/xml/kernel.xml.i
4 ./data/xml/kernels.xml.i
4 ./data/xml/kernels_header.xml.i
4 ./data/xml/head.xml.i
4 ./data/xml/fs.xml.i
4 ./data/xml/copyright.xml.i
4 ./data/xml/branch.xml.i
4 ./data/xml/archs.xml.i
4 ./data/server/utils.py.i
4 ./data/server/regexp.py.i
4 ./data/server/pgsql-kernel_group-branch.py.i
4 ./data/server/pgsql-fs_modules_pci.py.i
4 ./data/server/klive.tac.i
4 ./data/server/klive_pgsql_pwd.py.i
4 ./data/server/klive_error.py.i
4 ./data/server/iweb.py.i
4 ./data/server/.cvsignore.i
4 ./data/scripts/webalizer.sh.i
4 ./data/scripts/upload-klive.sh.i
4 ./data/scripts/upload-downloads.sh.i
4 ./data/scripts/stop_klive.sh.i
4 ./data/scripts/start_klive.sh.i
4 ./data/scripts/restart_klive.sh.i
4 ./data/scripts/remote_cleanup.sh.i
One other thing to consider is that tailor added only one additional
line for each changelog like this:
[project @ 2006-05-06 00:41:23 by andrea]
There are about 330 changesets, so that means 13K more of data due those
changelogs. Plus these lines should compress well, so it should be a lot
less than 13K of difference.
It's a pity I've a single file of this repo that contains a password
so I can't make it public. I guess I should regenerate the hg tree from
cvs after deleting the corresponding ,v file in cvs. Is it possible to
delete a .i file and get away with it with HG?
> Does your CVS repo have lots of branches?
Perhaps a few tags but no branches at all (I have never been confortable
with branches in cvs ;).
More information about the Mercurial
mailing list