Re: [bug-gawk] bug report on PROCINFO[,"READ_TIMEOUT"]

[Andrew J. Schorr]

On Mon, May 11, 2015 at 03:43:26PM +0300, Aharon Robbins wrote:
> > Why is READ_TIMEOUT actually useful?  I can't think of any usage cases
> > where it wouldn't be preferable to use select.  Maybe I'm missing something.
> 
> I don't remember the motivation or who first did that code.
> I suspect it was for interactive I/O, so that a program didn't
> sit prompting for input forever.  But I really don't remember.

Based on my email archive, it looks like John Haque wrote the initial
implementation and patch in October, 2011.  It was incorporated into the master
branch in Jan, 2012, and the complaints started in May, 2013. :-) As far as I
can tell from a quick glance, the original version in
d84926f0d651d5f55b1be3b760b572a50197593f had this same behavior of setting an
error if a timeout occurs.

It's great to be able to set a limit on how long to wait, but the inability
to recover from a timeout seems to me to render this feature pretty useless.
The new RETRY feature in the master branch fixes the issue, although I wonder
whether READ_TIMEOUT should automatically enable RETRY.  I guess changing
the behavior is not a good idea, although I wonder if anybody has ever
made good use of the READ_TIMEOUT feature.

Regards,
Andy