On 12/06/13 15:31, Philip Martin wrote:
"Bert Huijben"<[email protected]> writes:
-----Original Message-----
From: Eric Lubin [mailto:[email protected]]
Sent: donderdag 5 december 2013 21:17
To: [email protected]
Subject: [PATCH] fix for programmer error in path split text logic
[[[
* subversion/libsvn_subr/path.c
(svn_path_splitext): Fixed ncorrect pointer comparison to NULL to check
for end of string, instead of checking if the character itself was NULL
]]]
With that code unchanged for years / many Subversion version, I'm not sure
if we should really apply the behavior change now.
Perhaps we should just remove the check for a final dot.
The only difference in the output is that for an input "a/b." the
path_ext return value changes from a pool allocated "" to a static "".
I've added a test case and committed the patch.
Wouldn't last_dot[1] be more readable than (*(last_dot + 1) ?
- Martin
