i've been doing some fun pex performance investigations recently and am now looking into the impact of optionally using

multiprocessing

to add sources and resources in parallel when building a pex. the method of specifying sources and resources is a little different too, where the inputs are provided in a json file with checksums so they can be retrieved from a cache. i think adding sources and resources is definitely highly parallelizable, so it will be interesting to see how this works out performance-wise. my goal right now is just to make regenerating the

pants.pex

used for testing in the pants repo super fast, the next goal is to allow resolving sources/resources/requirements from a remote cache at pex bootstrap time so that redeploying a pex could only involve sending over a json file. i swear i will dump this piece by piece into subsequent PRs -- still figuring out whether this is worth doing at all right now.