it's not a matter of clang not being able to find the right dynamic libraries for its own execution, but being able to find

crti.o

when compiling executables from other object files. i have just figured out how to add multiple directories to clang's search path (including its own) and i believe i can do what i almost did a while ago, which is collect the search directories from gcc and clang with their

-print-search-dirs

option, then apply them when executing both compilers. gcc successfully finds

crti.o

when compiling `python_dist()`s, so i believe this will work now that i've figured out how to add multiple search dirs to clang. that being said, i'm actually not sure i buy not providing glibc once we have the ability to provide executables (which we won't, for a while). i was able to unskip the native subsystem test by removing the part where we try to create an executable, but with this technique i may be able to add it back