Re: os.select misbehavior with pipe

Ken Manheimer (klm@NIST.GOV)
Tue, 10 Jan 1995 11:13:16 -0500 (EST)

> Our replies to Sjoerd crossed each other. I hope mine explained the
> situation sufficiently (no, there is no way to tell whether there is
> buffered data -- you have to make the file unbuffered).

Yes, my questions were settled by your reply.

> Here's a patch to posixmodule.c that adds a third 'bufsize' parameter
> to fdopen() and popen().

I'm going to try to implement the workaround you suggested, just so i have
something to post that can be used in existing python versions. (I figure
i can use the 'fileno' file method to get the handle for the unbuffered
reads, with little impact on my existing code, and little change to
accomodate both the old and new python versions.)

Ken
ken.manheimer@nist.gov, 301 975-3539