• Steve Teale (8/8) Nov 17 2011 I've noticed this sort of thing mentioned a couple of times recently in
• Trass3r (6/12) Nov 17 2011 A long running D service could easily pile up a lot of memory.
• Andrew Gough (11/22) Nov 21 2011 Hi Steve,

Steve Teale <steve.teale britseyeview.com> writes:

I've noticed this sort of thing mentioned a couple of times recently in 
conjunction with CGI.

I did a D implementation of the AJP13 protocol some time ago. That is 
very easy to set up in Apache - you just follow the instructions for 
Tomcat. Then you can make your D app a service/daemon so it does not have 
to start up to serve a request.

I can dust it off if anyone wants to use it.

Steve

Trass3r <un known.com> writes:

Am 17.11.2011, 14:02 Uhr, schrieb Steve Teale  
<steve.teale britseyeview.com>:

 I've noticed this sort of thing mentioned a couple of times recently in
 conjunction with CGI.

 I did a D implementation of the AJP13 protocol some time ago. That is
 very easy to set up in Apache - you just follow the instructions for
 Tomcat. Then you can make your D app a service/daemon so it does not have
 to start up to serve a request.

A long running D service could easily pile up a lot of memory.
CGI doesn't have that particular problem.

I'd really be interested in some comparisons, CPU utilization, memory  
footprint, response time, impact of druntime initialization, ...

Andrew Gough <andrew goughy.org> writes:

On Thu, 17 Nov 2011 13:02:50 +0000 (UTC)
Steve Teale <steve.teale britseyeview.com> wrote:

 I've noticed this sort of thing mentioned a couple of times recently
 in conjunction with CGI.
 
 I did a D implementation of the AJP13 protocol some time ago. That is 
 very easy to set up in Apache - you just follow the instructions for 
 Tomcat. Then you can make your D app a service/daemon so it does not
 have to start up to serve a request.
 
 I can dust it off if anyone wants to use it.
 
 Steve

Hi Steve,

I'd be very interested to see the code as I've got an AJP13
implementation also.  I've been working on a HTTP implementation
using the also so the handler code is protocol independent but I'd be
like to compare approaches if that's ok.

-- 
Andrew Gough
M: 0408 596 656
andrew goughy.org