BOINC Combined Statistics

Does it make sense to specify only the url of the project stats folder, attempt to download db_dump.xml and use the names of the files it generated verses making a separate entry for each file name? Using a method like this could make it easier to bring new projects online especially when the project uses non standard file names.

It would involve making a more complicated script and require that the programs which use the stats files first check the db_dump.xml file for the actual file name used, but having a 'plain' url table with project 'nicknames' might make mangement of the various project sites easier.

Sure, that would make sense, but my shell programming skills are sub-par (give me C++ or Java anyday over shell scripting/perl). Not to mention the fetch script is 100% decoupled from the import process. The import process uses the file names and paths stored in the DB table, the fetch script does not. If someone wants to update the fetch script to use the DB table to get the projects (Active='Y' required) and use the db_dump.xml - I'd be happy to switch to it and import it in to the subversion repository.

The one catch tho, is I use wget for getting the files as it will only pull the file if it was updated, and it also preserves the file date/time.

Are you able to use php(5) from your shell script?

site only has php4