RE: [PATCH bisected regression] input_available_p() sometimes says 'no' when it should say 'yes'

[Chris Purvis]

Apologies all.  Nic and I did not realise that an internal email had been cross-posted to a public mail list (let alone one with strict email formatting rules!), and were having a hard time understanding the context for the emails we were receiving.

I have a certain amount of experience of asynchronous communication and protocol design: we aren't novices in this area.  One would hope in the kind of intra-machine protocol that we're using here that, if we know the sender is halted, there should be a way of clearing the contents of the channel so that the receiver can get hold of whatever has been put in to it.
Can tcflush() (or some similar API) be used to resolve our debugging scenario?

Regards,
Chris
-- 
Chris Purvis
Senior Development Manager 

Micro Focus

chris.purvis@xxxxxxxxxxxxxx
The Lawn, 22-30 Old Bath Road
Newbury, Berkshire, RG14 1QN, UK
Direct:    +44 1635 565282



This message has been scanned for malware by Websense. www.websense.com