On Thu, Dec 23, 2010 at 11:21:56AM -0800, Christopher Lee wrote: > the endless problems with dbm style interfaces that people run into on > different Python installs (especially Mac) persuade me that we need to take > action on the dbsqlite dbm-style interface to sqlite as we had previously > discussed. It seems like this would really solve the problem, which has > wasted so much of our time already. Looking at issue 3783 in the Python bugs > database, it looks like no further progress has occurred there, so if we want > action on this we will have to take it up ourselves. > > I took a look at the dbsqlite code, which suggested the following: > > - we would need to implement basic handling of the c / n / w / r mode flags. > > - how do you think we should test this prior to pushing it into a Pygr > maintenance release (such as 0.8.3)? > > One obvious initial test is to run the Pygr test suite using this on mac 10.6 > apple-supplied Python 2.6, which lacks any decent dbm style interface (only > dbm works, and it lacks any way of iterating...). Running the full Pygr test > suite using dbsqlite would at least test it in a wide variety of areas > including sequence databases, worldbase, etc. This would at least give us a > basic sense of whether it solves our relevant Pygr needs... > > What are your thoughts on this idea of trying to use sqlite as a dbm / shelve > replacement? It's been a long time since we've discussed it.
Hey Chris, good idea. since then I've used the dbsqlite code in two separate projects, tagnabbit and pony-build, whtout any hitches. While I don't remember the specific issues driving these patches, I made two modifications: https://github.com/ctb/tagnabbit/commit/26db7053f05f579d1f2e801336ad759ac3468ed8 https://github.com/ctb/pony-build/commit/baa93ff5f0b014e8b01b0a1f0941b5f2b947363f The latter patch is important for proper iterator behavior, I believe. The first patch is for good emacs support. The general persistence system in pygr is due for a radical simplification and overhaul, but that's probably too ambitious for now. I agree with your proposal for testing. I'm on vacation right now, which means I may or may not get a chance to work on this. Ping me privately if you start working on it so we don't duplicate work. Note, for the SequenceFileDB fix, screed may be a good option: see https://github.com/ctb/screed/blob/master/screed/tests/test_pygr_api.py for an example of how to use it. Briefly, from screed.pygr_api import ScreedSequenceDB should give you a drop-in replacement for SequenceFileDB, although you'll need to index the database separately; see https://github.com/ctb/screed/blob/master/doc/example.txt cheers, --titus -- You received this message because you are subscribed to the Google Groups "pygr-dev" group. To post to this group, send email to pygr-...@googlegroups.com. To unsubscribe from this group, send email to pygr-dev+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/pygr-dev?hl=en.
