For science, my list. --- * Portability. (What was this list called again?) * Related to portability: keeping up with the evolution of UNIXy APIs. We are a bit behind, e.g. the ...at() variants taking dirfds. Some socket stuff. pread, pwrite. * I think some of the List::Util and Scalar::Util utils really don't need to be in an extension. Lots of small utilities like this exist, in many cases we already have the C API, which just needs exposing. * The syscall API (especially for files and I/O, probably also for sockets, which are just another kind of I/O) needs virtualizing. (That is, one could trap open() and unlink() and read() and ...) This allows for many nice things like supporting "virtual filesystems", and nicer testing. (There's already something in the perltodo about this.) * Mersenne Twister in the core, or in a dist/ext. Yes, I have Python envy here. But that is where languages seem to be all standardizing. * The netlib-gdtoa library for portable and precise floating point input and output. Many I probably forget.Thread Previous | Thread Next