>I think the key here is that the shell should do what it takes to make
>things work correctly.

If by "things" you mean itself, then yes.  If it can't tolerate
O_NONBLOCK being on, then it should be responsible for turning
it off.

Generally it probably ought to be off, since the general Unix
paradigm is that reads are blocking if data is not available.
Specific things that want to poll are expected to turn that
feature on, and given "Death from the Skies" of Unix the managing
shell should probably keep turning it off, since the offender
can't be expected to always clean up after itself.

-- Jim




_______________________________________________
busybox mailing list
[email protected]
http://lists.busybox.net/mailman/listinfo/busybox

Reply via email to