Re: Subversion performance (issue #1429 et al)

"Sander Striker" <striker@apache.org> writes:

> In short, we need to look at svn_fs_is_dir and delta_proplists
> at the very least.

svn_fs_is_dir() is likely as tight as it can realistically get -- very
simple function, but the question it asks comes at a non-trivial cost,
thank you DAG filesystem. :-)

But here's the kicker -- the dir-delta (and, well, all filesystem
traversal code) could avoid calling svn_fs_is_dir() altogether if only
the svn_fs_dirent_t items reported the node's kind. This would add
cost to svn_fs_dir_entries(), though, so perhaps we could add a flag
to that function specifying whether or not we should populate a new
'kind' field of the svn_fs_dirent_t().

I'll take some time to see who all uses svn_fs_dir_entries(), and how
they use it, and determine if this change would be Goodness or not.
If Goodness, I'll code the change and commit.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sun Aug 17 16:53:53 2003