On Tue, Jan 21, 2014 at 02:32:57PM +0100, Daniel Willmann wrote:
> I'm not opposed, but I don't see how it would help anything. I doubt
> that whoever forgot to check the function signature will remember to
> compare the length against a define.
Maybe it is time for gerrit or at least a set-up for patchwork. I had
a look at your branch and stopped at the first commit message. :)
"Before this patch we just loop over the fd with read returning -1.."
==> looped (past tense)?
"EINTR is handled to not cause an error and because the socket is not
opened with O_NONBLOCK we don't need to check EAGAIN/EWOULDBLOCK."
I don't think it is true. osmo_fd_register will enable O_NONBLOCK on
the socket and I think we use that to register the fd.
the actual fixes look nice though. :)
