Stefan Fuhrmann wrote on Thu, Jun 02, 2011 at 00:08:16 +0200:
> On 01.06.2011 16:08, Julian Foad wrote:
> >On Sun, 2011-02-20, [email protected] wrote:
> >>Merge all changes (r1068695 - r1072516) from the
> >>integrate-stream-api-extensions.
> >>
> >>These patches add svn_stream_skip(), svn_stream_buffered()
> >>and svn_stream_supports_mark().
> >A belated review.
> >
> <snip>
...
> >The relevant question right now is: is there a clear benefit for this
> >being public in 1.7?  I think not, so I propose to make the "is
> >buffered" API private by renaming the relevant symbols to
> >"svn_stream__*".  Or move them into private/svn_io_private.h, as Stefan
> >just suggested.  Questions of semantics can then wait.
> Make it private. It simply felt a little odd to require
> every stream to implement a private API as well.
> 

Agreed, since the stream API is public it needs to work for streams
defined by third parties too.  I suppose we could implement this by
answering "FALSE" by default in _is_buffered() and _supports_mark().

> -- Stefan^2.
> 

Reply via email to