On 07.01.2013 15:52, C. Michael Pilato wrote: > On 01/07/2013 06:12 AM, Justin Erenkrantz wrote: >> On Sun, Jan 6, 2013 at 9:44 PM, C. Michael Pilato <cmpil...@collab.net >> <mailto:cmpil...@collab.net>> wrote: >> >> hosted elsewhere for them. The BDB backend (thanks to improvements to >> the >> Berkeley DB library itself) is much more stable today than it was when we >> first started this project, so it's quite possible that we don't hear >> noise >> >> >> That's quite surprising. My understanding from the Sleepycat/Oracle team >> way back when was that our core usage of BDB was wrong and would never be >> properly supported by them. Have they embraced multiple reader/writer >> processes now, or do they still advocate that having a single-process is the >> only Right Way(tm)? -- justin > Subversion's core usage of BDB arguably helped to advance the state of the > Berkeley DB art. Multi-process support has been officially part of the > Berkeley DB API since at least the 4.4 release (Branko will remember the > original DB_REGISTER work, I'm sure). And just few years ago I was even > having private conversations in which ex-Sleepycat folk were trying to urge > the Subversion project to further embrace Berkeley DB, specifically its > built-in database replication support. Single-process (via a brokering > daemon) is still the recommended approach, of course, but my understanding > is that it's no longer considered the Only Right Way.
Nevertheless the situation we have now is what it is. I see only two ways forward: stop supporting BDB (eventually), or invest serious effort into making it competitive. If we decide for the latter, it should be for good reasons; maintaining two back-ends is, as we know, a big investment. Personally I would rather spend that time doing sexy new things, such as the FSv2 API (getting rid of DAGs in the FS interface would be a good thing), better merging, replacing externals, etc. etc. -- Brane -- Branko Čibej Director of Subversion | WANdisco | www.wandisco.com
