On Aug 22, 2011, at 10:16 AM, "Cédric Krier" <[email protected]> wrote:
> Tryton was design to be run into many processes but the default implementation > only use threads. It is good for simple usage and on common hardware. > So even if we could run multiple instance of trytond, it required some > development to get a good way to dispatch requests to each processes. > > The first idea was to implement some process forking with multiprocessing but > it will require some re-work of the current BaserServer. > The second idea was to use existing framework to do the job for us and it > comes that wsgi is the common interface for such task in Python world for such > tasks (and also because of dropping Net-RPC and having only HTTP protocols). > > So the modivation was to provide an easy way to take benefit of the design of > trytond on large multi-core server (as Python can run on only one at a time). So implementation would be to configure Tryton to run under a wsgi-capable web server like Apache? It appears that lighttpd and nginx both work with wsgi apps but the support seems somewhat hacked together. Are there more pythonic ways of doing it? Is this also a step toward using Tryton as a web framework? As I recall reading somewhere, that has been done using werkzeug. -- [email protected] mailing list
