Github has the same problem for the same reasons - some languages are ignored because they're considered to be documentation instead of code. Regularly reported by users over there, never corrected; ... [More] Github considers it a bug if the automated analyser fails to correctly identify a language, but not a bug if a correctly-identified language is excluded by the system-wide configuration. The problem seems worse on Ohloh because these stats are a bigger part of what Ohloh does - and the Ohloh nutshell badge quotes statistics on two different definitions of code right next to each other in a misleading way: 77,564 lines of code [counting MetaPost]... mostly written in C [not counting MetaPost]. It'd make sense to me to simply count ALL languages unconditionally. Then there'd be no problem of figuring out which ones should or should not count. But I imagine that then some projects actually would end up being classified by the languages of their documentation instead of their code, and complain about it. The obvious thing to do is allow projects themselves to designate which languages and files should be counted. Ohloh already has a feature for excluding (but not for explicitly INCLUDING) files, partially addressing that. [Less]

My project's code base (see link) is 68% MetaPost and correctly detected as such. MetaPost appears at the top of the list on the analysis page, accounting for roughly 35000 of the roughly 50000 lines ... [More] in the project. However, the factoids say the project is Mostly written in C. Why? And can that be changed? My best guess is that maybe someone thought MetaPost code is in some way documentation rather than really real code, and the factoid generator only looks at a short list of real programming languages instead of the entire list that the line counter looks at. But if that was the rationale, it's simply incorrect. [Less]