Posted in Semantic Web, tagtriples on Jul 30th, 2007 No Comments »
I’m trying to work out if it’s possible to get the index searching performance I want using a disk-backed store, or whether I need to focus on optimising the indexes to fit totally in memory. The problem is that the optimisation strategies are somewhat different:
Storing indexes on disk:
- Increase redundancy, trading space for better locality […]
Posted in Semantic Web, tagtriples, data, indexing on Jul 26th, 2007 3 Comments »
The new triplestore is coming along. It can do substring text searches (using a suffix array) and has a basic relational query engine. It doesn’t optimise the query plans yet, but if you enter the queries in a good order (most selective clauses first) then you get good performance.
A few things have changed in my […]
Posted in General, programming, scheme, gambit on Jul 25th, 2007 1 Comment »
Gambit scheme lacks a profiler that can profile scheme with embedded C code. (There’s statprof, but unfortunately it doesn’t profile embeded C). I needed to do this pretty desperately for my triple indexing stuff so I’ve written a simple macro which takes wall clock timings of functions and accumulates them.
You replace ‘define’ with ‘define-timed’ […]
Posted in General, programming on Jul 23rd, 2007 3 Comments »
I’ve been struggling to motivate myself recently. I spent a lot of my spare time in my twenties programming free stuff after work with no trouble at all, but now in my thirties the energy and enthusiasm seems a little harder to come by. Plus Claire and I are expecting a baby any day now […]
