Had an unwilling flashback to early 2000-s style computing earlier today when I tried to install a package (pycddlib) for #python. Due to some weird dependency bug in the most recent versions of python, this ended up taking two hours of downloading various bits and pieces of software (in no particular order: cython; .NET SDK; gmp; MPIR; VSCode Solution Explorer; vcpkg (attempted, but abandoned); CMake (attempted, but abandoned); MSYS2; a single file longintrepr.h (manually downloaded from a github repo); several unrelated github projects that shared some of the dependencies (but did not ultimately advance the resolution, probably due to the lack of fundamental understanding of what I was actually doing at some stages of the process); and several manual compilations and "make"s, updates of PATH environment variables and reboots), as well as multiple google searches (mostly to stackexchange sites and similar) and consultations with AI, before finally succeeding (after three partial successes in which the immediate error message was resolved, only to encounter a fresh error message). This sort of extremely tedious problem-solving is a task which I will be very happy to see become a thing of the past when personalized AI-empowered digital assistants become commonplace!