[Catalyst-dev] Catalyst test suite

[Andy Grundman]

The POE engine also has a blocking problem, a really big one in fact,  
and is the reason I gave up on it.  What you want to do is currently  
not possible, a Catalyst request will generally block the whole  
process until it's finished executing.  The POE engine tried to use  
some tricks to get around parts of the blocking, but the core of your  
app code, and things such as database queries, are still going to block.

On May 6, 2008, at 1:49 PM, Yuri Shtil wrote:

> I take it back.
>
> As I said, I want to be able to integrate a Catalyst engine into an  
> events loop and was contemplating to write a new engine.
> I foresee a need for load balancing and since the fastcgi engine.  
> However I wanted to start with something that is  simple working and  
> supported.
> I guess I would go with HTTP-POE in spite the fact it is not  
> supported. The prefork engine will have the same blocking problem in  
> each forked subprocess, I guess.
>
> Andy Grundman wrote:
>> Oh for some reason I thought you were writing a new engine.  What  
>> problems are you having with the FastCGI engine?  There are already  
>> test scripts for testing lighttpd and Apache 1/2 with FastCGI.
>>
>> On May 6, 2008, at 1:30 PM, Yuri Shtil wrote:
>>
>>> Catalyst::Engine::FastCGI <http://www-homes.juniper.net/%7Eyuris/perldoc/lib/Catalyst/Engine/FastCGI.html 
>>> >
>>>
>>> Andy Grundman wrote:
>>>> What sort of engine are you working on?
>>>>
>>>> On May 6, 2008, at 1:13 PM, Yuri Shtil wrote:
>>>>
>>>>> I guess my question then would be:
>>>>> - how do I handle actions that take a while to execute, for  
>>>>> example if I need to talk to an another process that can take a  
>>>>> while to respond, is there a way to integrate the Catalyst  
>>>>> server into an external events loop be it POE, EV or something  
>>>>> else.
>>>>>
>>>>> Andy Grundman wrote:
>>>>>> I guess we weren't testing the same thing... I was running t/ 
>>>>>> optional_http-server.t not the POE test.  BTW, the POE engine  
>>>>>> is no longer supported (by me anyway), it's been deprecated in  
>>>>>> favor of the HTTP::Prefork engine.
>>>>
>>