On Tue, Jun 11, 2013 at 3:56 PM, Bert Huijben <b...@qqmail.nl> wrote: > > >> Hi, >> >> Currently Subversion client layer creates new RA session for every >> svn_client_* call. Even more: for some operations like >> svn_client_merge() it creates 10-15 RA sessions. Each session creation >> takes significant amount of time: TCP connection, SSL handshake, >> authentication and initial handshake. It easily could take several >> seconds over WAN. >> >> To solve this problem I propose to introduce RA context abstraction >> layer for managing set of RA sessions for different repositories and >> reuse them if possible. In the first version I made this layer >> libsvn_client private, but in we can make it part of ra-loader in >> future. >> >> The patch is attached. It passes all tests of course. I'm very >> interested for feedback. Proposed solution should make our svn merge >> code much easy to maintain and significantly faster. >> >> [[[ >> Introduce repository access abstraction layer for managing RA session for >> different repositories. >> >> * subversion/include/private/svn_client_private.h >> (svn_client__ra_session_release): New. >> (svn_client__ra_session_from_path2): Document that created session will >> be >> automatically returned back to RA session cache on pool cleanup. >> >> * subversion/libsvn_client/ra.c >> (): Include ra_ctx.h. >> (svn_client__open_ra_session_internal): Use >> svn_client__ra_session_open(). >> (svn_client__ra_session_release): New. Wrapper around >> svn_client__ra_ctx_release_session() >> >> * subversion/include/svn_client.h >> (svn_client_ctx_t): Add RA_CTX member. >> >> * subversion/libsvn_client/ctx.c >> (): Include ra_ctx.h. >> (svn_client_create_context2): Initialize RA_CTX. > > I haven't looked at the code yet, but is this going to work for clients that > keep a > svn_client_context_t instance for hours/days/weeks? > All that time reusing it for different operations every now and then. > (One step further: How would this be affected by configuration and/or > authorization baton changes when there are sessions cached) > > Existing TCP connections (w/c)ould be broken on the next usage, kerberos > cookies could have expired, etc. etc., especially with ra_svn, but maybe also > with ra_serf where an existing connection can be reused in some cases. > It should be working fine, the only problem is ra_svn. But it should be relatively easy to solve. RA session should be restartable anyway.
-- Ivan Zhakov CTO | VisualSVN | http://www.visualsvn.com
