Comments on: Using custom functions with SQLAlchemy and SQLite http://ginstrom.com/scribbles/2008/08/10/using-custom-functions-with-sqlalchemy-and-sqlite/ Random scribbling about programming, translation, and Japan Fri, 03 Feb 2012 09:05:20 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Ryan Ginstrom http://ginstrom.com/scribbles/2008/08/10/using-custom-functions-with-sqlalchemy-and-sqlite/comment-page-1/#comment-267 Ryan Ginstrom Thu, 14 Aug 2008 13:23:30 +0000 http://ginstrom.com/scribbles/2008/08/10/using-custom-functions-with-sqlalchemy-and-sqlite/#comment-267 I probably didn't explain the algorithm well enough. You've got to iterate through each record in your glossary, retrieving all the records with gloss_score(query) gte minscore. SQLite iterates through lots of records much faster than pure Python. I probably didn’t explain the algorithm well enough. You’ve got to iterate through each record in your glossary, retrieving all the records with gloss_score(query) gte minscore. SQLite iterates through lots of records much faster than pure Python.

]]> By: Amir Helzer http://ginstrom.com/scribbles/2008/08/10/using-custom-functions-with-sqlalchemy-and-sqlite/comment-page-1/#comment-268 Amir Helzer Thu, 14 Aug 2008 11:53:13 +0000 http://ginstrom.com/scribbles/2008/08/10/using-custom-functions-with-sqlalchemy-and-sqlite/#comment-268 I probably didn't understand your original algorithm. What I meant is that if there's a long array and you're implicitly searching it (like array.count()), you're getting execution time that's linear to the array's size. I probably didn’t understand your original algorithm. What I meant is that if there’s a long array and you’re implicitly searching it (like array.count()), you’re getting execution time that’s linear to the array’s size.

]]> By: Ryan Ginstrom http://ginstrom.com/scribbles/2008/08/10/using-custom-functions-with-sqlalchemy-and-sqlite/comment-page-1/#comment-266 Ryan Ginstrom Wed, 13 Aug 2008 03:38:08 +0000 http://ginstrom.com/scribbles/2008/08/10/using-custom-functions-with-sqlalchemy-and-sqlite/#comment-266 @Amir I'm not sure whether a dictionary would gain anything, because this is an iteration problem, not a search problem. You've still got to iterate through every entry in your database, because your search query is new each time. The dictionary would speed things up if you could leverage the hash function. @Amir

I’m not sure whether a dictionary would gain anything, because this is an iteration problem, not a search problem. You’ve still got to iterate through every entry in your database, because your search query is new each time. The dictionary would speed things up if you could leverage the hash function.

]]> By: Amir Helzer http://ginstrom.com/scribbles/2008/08/10/using-custom-functions-with-sqlalchemy-and-sqlite/comment-page-1/#comment-265 Amir Helzer Tue, 12 Aug 2008 14:30:27 +0000 http://ginstrom.com/scribbles/2008/08/10/using-custom-functions-with-sqlalchemy-and-sqlite/#comment-265 Hi Ryan, Searched through Python arrays will take order(n) time, no matter what you do. Psyco will not help at that at all. I like your SQLAlchemy solution. Just for fun, if you compare its performance against a Python dictionary (instead of array) or the same size, you'll see completely different results. Searching through a dictionary takes order(log-n) time. Inserting to the dictionary would be slower - but not a killer. Amir Hi Ryan,

Searched through Python arrays will take order(n) time, no matter what you do. Psyco will not help at that at all.

I like your SQLAlchemy solution. Just for fun, if you compare its performance against a Python dictionary (instead of array) or the same size, you’ll see completely different results.

Searching through a dictionary takes order(log-n) time. Inserting to the dictionary would be slower – but not a killer.

Amir

]]>