On Thu, 2011-06-02 at 11:49 +0100, Julian Foad wrote:
> On Thu, 2011-06-02, Stefan Fuhrmann wrote:
> > On 01.06.2011 17:06, Julian Foad wrote:
> > >> svn_error_t *
> > >> svn_stream_skip(svn_stream_t *stream, apr_size_t *len)
> > >> {
> > >> SVN_ERR_ASSERT(stream->skip_fn != NULL);
> > > This should call the default skip implementation (which reads and
> > > discards) when a stream doesn't implement its own skip function. That
> > > way, 'skip' will be available on any generic stream, which is better
> > > than the caller having to know (either implicitly or by calling a test
> > > function) whether the stream has a skip function installed.
> > Agreed.
> > >> return stream->skip_fn(stream->baton, len);
> > >> }
> > >
> > >> static svn_error_t *
> > >> skip_handler_empty(void *baton, apr_size_t *count)
> > >> {
> > >> *count = 0;
> > >> return SVN_NO_ERROR;
> > >> }
> > > I thought a skip handler was supposed to return an error when asked to
> > > read past EOF. This one doesn't. Anyway, if we make the above change
> > > then we can get rid of this handler; the empty stream doesn't need one.
> > +1. This function is unlikely to be called for an empty
> > stream anyways.
> >
> > >> static svn_boolean_t
> > >> is_buffered_handler_empty(void *baton)
> > >> {
> > >> return FALSE;
> > >> }
> > > We don't need this handler, because there's a default for streams that
> > > don't supply a handler.
> > Back then, I had just discovered that translated streams (?)
> > did not provide a specialized skip() implementation. That
> > prompted me to look for any "missing" handler.
> >
> > But this one is indeed not needed.
>
> Done in r1130490: "Change svn_stream_skip() so that if no skip handler
> is provided it will still work by reading from the stream, and so remove
> three trivial handlers."
And removed translated_stream_skip(), which was a copy of
skip_default_handler(), in r1130498.
- Julian
Received on 2011-06-02 13:19:07 CEST