Re: svn commit: r1561387 - in /subversion/trunk/subversion: bindings/javahl/native/ include/ libsvn_ra_svn/ svnserve/ tests/libsvn_ra/

Following up from irc:
> [23:23] <Bert> brane: Any ideas for a better api? One problem is that
> ra_svn already abuses a stream in the same way internally... so
> exposing that is not really helping.
> [23:24] <Bert> brane: And inventing yet another stream api... Yuck.
> [23:25] <Bert> (but if we do introduce a new stream api... we should
> add proper scratch pools)
> [23:26] <Bert> brane: The stream api is pretty much what you would
> need to use blocking tunnel libraries...

Agreed on all of the above. The best solution, IMO would be to change
the stream API constraints, so that only a zero-length read represents
an end-of-file. The current constraint looks like a clear design bug to me.

But I'm not sure how to do that without rev'ing the whole stream API ... :(

— Brane

-- 
Branko Čibej | Director of Subversion
WANdisco // Non-Stop Data
e. brane_at_wandisco.com