On 2012-05-15 09:56, Mark Marsella wrote: > Mongrel2 isn't a language so pep is overkill. I think Mongrel should be > kept small and work should be focused on optimising and resolving > issues, rather than adding loads of new features. As long as it can take > http hand it to zmq then it's done it's job, and can be extended that way.
The full PEP is probably overkill, but a bit of formalism when we want to had new stuff would not be bad. For example we have things waiting for a decision: - UPPER_CASE headers for trusted headers from Mongrel2 and lower case for the others (overwritting the case of the client). - answer as tnetstring to provide more information (like error code on http), close connection after the answer, etc. - is the approach of storing locally to the webserver the body of a large request the best? Can it be stored locally, sent in one go later or served over an internal HTTP address? - HTTP chunked encoded from the client? Should we support it? How? These are issues where a bit of technical discussion is required and having a place where we can store formalized plan/arguments would make things easier for everybody. I am not talking about heavy workflow to validate proposal, more a virtual table where we sit, discuss and at the end we put down on "paper" the reasons, this for the hard parts. By the way Mongrel2 is not HTTP only, it is doing HTTP, Websocket, jssocket, xml. Regards, loïc > On Tuesday, May 15, 2012, Loic d'Anterroches wrote: > > Hello, > > as Zed is getting more fun writing books than coding at the moment, we > happy users and developers of/around Mongrel2 need to find a way > forward. > > In my personal/company case, Mongrel2 is a core part of my work, this is > why I need it to succeed. > > Practically, this means that: > > - I can setup a dedicated server for the project, I can take care of the > sysadmin and let trusted users "deploy" on it. > - I can put money where my mouth is to support the development. > > The second part is of course how to coordinate and how to select what > goes in Mongrel2 and what stays out. Maybe we can follow the Python PEP > approach for the parts requiring a bit of discussion. This is working > very well. > > In any cases, you can count on me to support the project for the years > to come and I am sure that after a while, Zed will come back to have fun > with the project. > > loïc > -- Dr Loïc d'Anterroches Founder Céondo Ltd w: www.ceondo.com | e: [email protected] t: +44 (0)207 183 0016 | f: +44 (0)207 183 0124 Céondo Ltd Dalton House 60 Windsor Avenue London SW19 2RR / United Kingdom
