>>115Maybe later, the benchmarks aren't a high priority, I've been doing manual performance regression testing periodically, I'm in a good state.
>>120To a degree, yes, it's worth it in this case. That change does make things somewhat simpler from the library maintainer's standpoint, it removes the cognitive overhead of having to think about making sure I'm not fucking something up with the intptr_t -> int casting. Leaving it as is makes it easier for someone to pass in a malformed allocid, and complicates the error checking. As far as losing the type-checking for users, it's not that big of a deal, just use the typedef and you'll be okay.
>>122Yes, they do, but they're needed for Travis CI (or pretty much any public continuous integration service). I could put them in a different branch, but then I'm not getting as continuous of integrations if I always have to merge from master into another branch or vice versa.