requests.{get,post} timeout

[dieter]

Chris Angelico <rosuav at gmail.com> writes:
> ...
> That looks like an exception to me. Not a "process is now terminated".
> That's what happened when I pressed Ctrl-C (the IP address was
> deliberately picked as one that doesn't currently exist on my network,
> so it took time).

What Jon argues about: signals are delivered to Python's main thread;
if a thread is informed (e.g. via a signal induced exception) that
a request (running in a different thread) should terminate, he needs
a way to make the different thread do that.


You may have argued before that in case of a signal, the request
fails anyway due to an EINTR exception from the IO library.

This may no longer work. Long ago, I have often been plagued
by such EINTR exceptions, and I have wished heavily that in those
cases the IO operation should be automatically resumed. In recent time,
I have no longer seen such exceptions - and I concluded that my wish
has been fulfilled (at least for many signal types).