On Mon, Feb 07, 2011 at 12:43:41PM +0100, Stefan Sperling wrote: > On Mon, Feb 07, 2011 at 10:30:15AM +0100, Branko Čibej wrote: > > On 07.02.2011 02:52, Branko Čibej wrote: > > > Note that I didn't run the tests with this patch, so I'm not claiming it > > > to be bug-free. In fact there's one bug in the recursive proplist of the > > > WC root, where the root props show up twice -- there's an issue with the > > > filter in the query there. But it's proof-of-concept after all. > > > > Well, I fixed that bug and the new patch is attached. I ran > > prop_tests.py, and did get one failure (see below); could be a > > difference in the output (the results of a recursive proplist are > > ordered now), but I don't plan to investigate that; whoever takes this > > patch and runs with it is welcome to do so. :) > > Thanks a lot! I'll take a look.
I like this patch a lot and think we should commit it after fixing the test failure. It seems like temporary tables are the answer to this problem. I expected this to be too slow, but it's very fast. Where is the temporary table stored? Is it back by a file or memory? If backed by memory, do we have to worry about memory consumption for large working copies?