W3C home > Mailing lists > Public > www-lib@w3.org > April to June 2001

Re: help wanted! key words : event loop, post, sockets, timers, p endi ng requests

From: Michel Philip <mphilip@infovista.com>
Date: Fri, 22 Jun 2001 11:03:12 -0400 (EDT)
Message-ID: <F520B214418AD4119F7000508BD90CC294B8D1@hq01sr02.infovista.com>
To: "'www-lib@w3.org'" <www-lib@w3.org>
> Hello all..
> i surely hope someone can point out what i need to do here...
> here goes:
> When sending a number of POST requests to an http server (thanks Aileen
> your example !!), this works exactly 33 times, after which i get the
> following messages in a debug session:
> Net Object.. starting request 140276200 (retry=1) with net object
> HTTP........ Looking for `
> HTHost parse Looking up `' on port 9004
> Host info... Found Host 1400663c0 with no active channel
> Host connect Grabbing lock on Host 1400663c0 with 1400c2500
> Host info... Added Net 1400c2500 (request 140276200) as pending on pending
> Host 1400663c0, 0 requests made, 0 requests in pipe, 1 pending
> HTDoConnect. Pending...
> HTHost 1400663c0 going to state TCP_CHANNEL.
> Funny thing is, when i quit my program and run it again, 
> i get exactly the same behavior (or behaviour :-))

After only 33 times that's a rather good condition to catch the bug.

> After which exactly nothing happens any more. 
> My guess is that, since i use my own eventloop, 
> i need to register some extra callback functions in order
> to handle pending requests. 
> Am I right ? if so, what do i need to do ?and if
> not, could someone point me towards the right direction ?
> It seems that whatever timer mechanism i use, those don't have any effect
> pending requests, but i suppose that's the way it's supposed to be.

I wonder... 
Maybe nothing happens anymore because the library is inside an infinite
Check the CPU usage before killing.

> Anybody any clue ?

The number of messages you sent is too short.
Previous messages concerning the same host/channel
would be interesting to try understood why 
did you get the messages
   "... with no active channel"
   "... Grabbing lock ..."

If the lib effectively went in an infinite loop (one containing no
and if you are running under NT then maybe you didn't get the last message 

I don't know if the stderr is auto flushed under NT.

Received on Friday, 22 June 2001 11:38:45 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:33:54 UTC