On Sunday, April 7, 2013 1:51:11 PM UTC+2, Arnon Marcus wrote:
>
> As for the Tornado-web-socket example, I looked at the code, and couldn't 
> figure out how this would work in a production environment...
> Does it spawn a separate python interpreter for Tornado?
>
 
yes, it's very clear that there are two separate shells.

If so, how does it meld with we2py's controllers? It is unclear how this 
> works...
>

none, effectively.  it's a separate process that handles just the messaging 
part..... usually the user's browsers (trough whatever you use, usually a 
javascript component) asks for messages *from it *or send them *to it* . 
You can also send messages within your web2py code .....

It's meant to handle all the "messaging" part, leaving the web2py working 
in the "usual" way.
 

> What serves the Tornado web-app in production? Apache? How?
>

Tornado usually "stands" by itself. Gunicorn too. Apache doesn't handle 
well long-standing connection, usually the way to go if you need something 
in front is using nginx.
 

>
> As for the comet file - I can't find it - it seems it no longer exist in 
> the new version of the web2py source-code....
>

???? comet was referred only in the very first implementation.... web2py.js 
holds a websocket component.
 

>
> As for running web2py via gEvent - how should one deploy this in 
> production?
>

with anyserver.py
 

> Can it work with Apache the same way the wsgi-handler does?
>
 
Nope. I'm not sure you can find any gevented "something" running behind 
apache.
 

> Does it require/suggest a "gEvent"ed uWSGI under NginX ?
>
 
If you want to run a gevented web2py, *it's another story alltogether*, 
that's not "pertinent" to what websocket_messaging.py is.
BTW, you'd need to "adopt" gevent-friendly libraries, etc etc etc if you 
don't want to rely just on monkey patching, and code all your app with 
gevent-like statements if you want to "exploit" the real potential of 
gevent.

 

> This is all very bewildering...
> The documentation is very lacking...
>

I think you'd need to identify correctly what are your requirements and 
stop thinking that all of what you want can be achieved in a single-process.
Requirements for "threading-like" webserver, "forked" ones can be (with few 
limits) assumed as "equal" regarding the choosen programming style, but 
long-standing connections are very different.
That's why the messaging part is abstracted away in a lot of 
implementation: you do what needs to be done in your "normal" environment 
and leverage the "messaging" with another "component", maximizing the 
usefulness of an MVC framework and leaving the ultra-specific messaging 
implementation on something external alltogether, that does just what it's 
meant for.


-- 

--- 
You received this message because you are subscribed to the Google Groups 
"web2py-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to web2py+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.


Reply via email to