Hi, On Dec 10, 2007 9:12 PM, Esteban Franqueiro <[EMAIL PROTECTED]> wrote: > >> What about JCR-314, 935 (which you mentioned) and 1148, issues related > >> with the > >> fine grained locking? We've been testing the fix I proposed for JCR-1148 > >> in the > >> JCR-935 comments for some time now and haven't had any NPE due to it. > > > > I'm not sure how ready we are to switch to fine grained locking as the > > default SISM locking strategy. From a feature perspective it would be > > really nice, but I'm a bit worried about stability especially since we > > finally seem to have overcome the infamous concurrent versioning > > problems. > > I'm not saying it has to be the default, but it could be optional. I have a > patch for 1.4 that adds > a lockingStrategy attribute to the repository config (per repository), and it > has a few tweaks in > it. I can upload it for review. Our testing with 1.3.1 + FGL shows a > tree-fold improvement, so I > think it's worth to further develop this feature.
Agreed. I think we should do this, and look at resolving the issues you mentioned for Jackrabbit 1.4. It sounds like things are stable enough with your suggested fixes. What do others think? If it's a configuration option, people can always fall back to the current behaviour, and in case problems do come up with FGL, we can always make 1.4.x patch releases to fix the issues. BR, Jukka Zitting
