Bert Huijben wrote:
> Javahl always calls status with an abspath target as its conversion to
> a c string does that.
OK, thanks. In r1139008 I've fixed the doc string and also documented a
promise that svn_client_status5() will send an abspath to the callback
if given an abspath.
- Julian
> Bert Huijben (Cell phone) From: Julian Foad
> Sent: donderdag 23 juni 2011 18:48
> To: dev_at_subversion.apache.org; Hyrum Wright
> Subject: svn_client_status_func_t callback - Java binding may expect an
> abspath
> Hyrum or anyone,
>
> svn_client_status_func_t is declared as
>
> /** A callback for reporting a @a status about @a local_abspath.
> ... */
> typedef svn_error_t *(*svn_client_status_func_t)(
> void *baton,
> const char *path,
> const svn_client_status_t *status,
> apr_pool_t *scratch_pool);
>
> but its "path" parameter is not called "local_abspath" and is not
> guaranteed to be an abspath, which is evident from the implementation.
>
> I am satisfied that all but one of the current callbacks is happy with a
> non-absolute path. The only one I can't validate is the JavaHL binding:
>
> SVNClient::status() in SVNClient.cpp
>
> where, according to the parameter names, it expects an abspath.
>
> Can someone better able to follow the Java code check this?
>
> Then I think we just need to adjust the doc string to say something like
> "path, which may be absolute or relative", and rename the JavaHL
> parameters to "path".
>
> - Julian
Received on 2011-06-23 19:33:12 CEST