Well, if you look at the benchmarks, several things stand out. 1. All the WSGI servers are pre-loaded. That means the code is already pre-compiled and loaded in ram, ready to run. Web2py doesn't quite work that way, because each request loads the database, does migrations if necessary, and inserts a lot of objects into the scope.
2. They are all using gunicorn, which spawns about 4 workers on startup. We could certainly run Web2py in WSGI mode with gunicorn, but it wouldn't make much a difference. I believe if we are going to support sessions, they should be off by default and use a decorator @session_enabled or something like that. However, that may go against Web2py principles. 3. Those benchmarks are not development type setups. If we are going to compare benchmarks, we'll need to make sure that our configuration matches theirs. 4. Web2py was meant to be a teaching tool. However, as it is now, what it's teaching is years old. Now everything needs to be WSGI so you can run it on a PAAS such as Heroku, Iron.io, Azure, etc. It's good to know how we compare with performance, but at the same time, I think we need to be realistic. Web2py needs to be re-architectured if we're going to see wider adoption in the future. On Wednesday, May 8, 2013 5:31:19 AM UTC-7, Alfonso de la Guarda Reyes wrote: > > Hi, > > How we can improve web2py to match the fastest python web frameworks > performance? > > http://www.techempower.com/benchmarks/#section=intro > > Surely, with many brains we can offer some method to improve that! > > Saludos, > > -------------------------------- > Alfonso de la Guarda > Twitter: @alfonsodg > Redes sociales: alfonsodg > Telef. 991935157 > 1024D/B23B24A4 > 5469 ED92 75A3 BBDB FD6B 58A5 54A1 851D B23B 24A4 > -- --- 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.
