Philip Martin <philip@codematters.co.uk> writes:
> Rev 3113 was when I changed the path stuff. As far as I can tell by
> looking at the rev 3112 code, I didn't change the behaviour of either
> svn_path_split("/"), which always returned "/", or that of
> svn_path_is_empty("/"), which always returned FALSE. Thus I think
> these bugs may have always been present, but nobody triggered them.
>
> Rev 3138 fixed a similar bug to 3150 (and rev 3113 itself also fixed
> one). I identified two suspicious loops around the time rev 3138 was
> being discussed, you have now fixed one, the other was in
> ok_to_bump_path().
Okay, thanks.
But there's still a bit of an abstraction failure, in that the code in
split_url.c "knows" that "" is the empty path, yet path.c wavers
between using a #define for that and just hardcoding it?
I feel like we could be more clean about this... If "" is the only
possible empty path, then really, why do we need svn_path_is_empty()
at all? We could just test path[0] directly every time.
-K
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Sep 10 17:14:12 2002