I would avoid having X use cthreads. It might assume that stdio will be
multithreaded, which ours isn't. I'd just let X be all single-threaded
until we have pthreads support implemented on the Hurd.
Check the code that is conditionalized on HasShm, and see how it handles
errors from the calls. If it handles errors in general well, or handles
ENOSYS specifically, then defining HasShm to YES should be ok. But don't
worry about it too much; other things affecting the X configuration are
likely to change before shm gets implemented.
The value for Malloc0ReturnsNull should be the same as for glibc on Linux.
I don't know any reason you'd want NO_ALLOCA, but whatever the rationale
behind that one way or another, the issue is most likely the same on Hurd
and on Linux.
Hurd does have the `poll' call as well as `select', so you could define
HasPoll to YES. (It's not particuarly better to use one or the other on
the Hurd; both can handle an unlimited number of file descriptors.)
