Re: Technical question wrt lua_pcall() and errors

[Marc Balmer <marc@...>]

Am 18.01.11 14:40, schrieb Marc Balmer:
> Am 18.01.11 14:06, schrieb Marc Balmer:
>> Am 18.01.11 13:41, schrieb Romulo:
>>> It might not be the case, but it happened to me that the httpd's error
>>> handler made a second request to the fastcgi url, thus, repeating the
>>> response.
>>> I'm sorry to don't quite remember it, but I guess it was the
>>> favicon.ico (I know, the browser is the one to blame, but still..)
>>> Perhaps you could print the request URI on top of each fcgi accept loop.
>>>
>>
>> I think the reason must be something different.  It only happens, and
>> really only, if the Lua function called with pcall() generates an error.
> 
> I can now tell that actually two exchanges between the webserver
> (lighttpd) and the FastCGI handler happen (verified with tcpdump), but
> only if the Lua code called with pcall generates an error.
> 
> I am more and more confused by this...

When my confusion reached the maximum I sat back and had a nice glass of
Balvenie Single Barrel, 48.5%, and began to think.

What I saw was just not possible.  And thus it was.  It turns out the
reason for all this was in a total different part of the system.

Many thanks for all that tried to help!