Re: FastCGI, Comet and Client Link failure

Hi Jay,

I start a new comet connection after I accept a FCGX_Accept_r returns a FCGX_Request with a URL of "http://<my server>/new_comet_connection_please". I can then send chunks down this connection using FCGX_FPrintF. I do not call FCGX_Finish_r for this request because I want to leave the HTTP transaction open for further data.

So fastCGI is giving me access to a new HTTP transaction and the FCGXFPrintF function allows my fastcgi application to send chunks to the web server (Abyss at present) to pass to the remote Client. So far so good. My problem is that at some point the TCP connection to the Client may die and I don't at present have any way of detecting this.

Cheers



________________________________
From: Jay Sprenkle [mailto:[email protected]]
Sent: 14 January 2010 15:24
To: Howard May
Cc: fastcgi-developers-xGejAJT2w6xVgU18Zptdi0EOCMrvLtNR@public.gmane.org
Subject: Re: [FASTCGI] FastCGI, Comet and Client Link failure

fastcgi is a transactional style application. If you want to run a persistent program you probably want the fastcgi application to communicate requests to, and retrieve status from, the daemon/service that implements your comet connection.

On Thu, Jan 14, 2010 at 6:23 AM, Howard May <Howard.May-3Vf4qVogE1JWk0Htik3J/[email protected]<mailto:Howard.May-3Vf4qVogE1JWk0Htik3J/[email protected]>> wrote:
Consider a simple fastCGI application with a comet connection to a remote Client. The application is periodically sending chunks to the Client but at some point the Client connection fails. How and when does the fastCGI application learn about this?

_______________________________________________
FastCGI-developers mailing list
FastCGI-developers-xGejAJT2w6xVgU18Zptdi0EOCMrvLtNR@public.gmane.org
http://mailman.pins.net/mailman/listinfo.cgi/fastcgi-developers