Hello,
I started using semmlecode on a real project and I need to analyze binaries: ie, I want to run queries on bytecode jars, zips and the like. Is this possible ?

Thanks a lot

Arnaud Bailly

Hi Arnaud,

By default, SemmleCode only stores those elements from libraries in the database that are directly referenced from the source. To store *everything* in the database, go to Preferences -> SemmleCode and tick the "parse libraries" option. You'll get what you ask for - on most projects, this will add millions of tuples to the database.

The fromSource() predicate is used a lot in metrics, because that is typically what is required - the definition of Lakos level, for example, explicitly makes a distinction between elements in source and libraries, stipulating that library elements do not have a level.

Queries should just work on libraries if you enable "parse libraries" as indicated above. They may take a long time, however, since you'll be processing so much data. Do let us know if you find particular queries take too long  - we've got some further optimisations up our sleeves...

Hope this helps,

-Oege

Hi Arnaud,

Yes - the facility is admittedly not user-friendly, but you can select (and de-select) libraries while creating working sets in the population dialog. We'll look into improving this - thanks for the suggestion!

-Oege