On Wed, Apr 23, 2014 at 6:02 PM, Emmanuel Lécharny <elecha...@gmail.com>wrote:
> Le 4/23/14 5:09 PM, Jeff MAURY a écrit : > > Hello, > > > > after a busy Devoxx France, I will be able to spend more time on MINA in > > the next weeks. > > I would like to share some thoughts with you: > > > > > > - HTTP: as I was presenting a talk at Devoxx about Web stacks > > performances, I was really upset by the fact that most of them are > using > > Netty as the underlying HTTP engine. I understood the reason: our HTTP > > codec is broken in many ways related to server development: no ZIP > support, > > no application/x-www-form-urlencoded or multipart/form-data support. > So our > > implementation is leaving too much work on the web stack > implementation. > > Not clear on my side for the client part. So I suggest complete > reworking > > the HTTP layer > > +1. > > > - Benchmarks; we have started working on this issue but the bad part > is > > that our tests are not distributed so the TCP layer is not really > used and > > may impact the result. I propose to set up a new environment based on > > Docker so that the server side can run automatically from Maven on a > remote > > machine. We should also try to keep the results as part of the build > files. > > As ASF does not seem to provide a distributed environment, > We can ask for machines where we can run some tests. > Can it be automated or is it a case by case ? > > OTOH, I have 4 Raspberry PI that could be used as injectors if needed. > > > I propose to run > > the build on my personal machines (MacBook + Ubuntu) and to archive > the > > results on my ASF home space > > > > - SSL: We've refactored the SSL process to be more event oriented, > but I > > think we should complete it, mainly related to rehandshake > > Absolutely. > > > > > > WDYT ? > > That's a perfect roadmap ! > > On my side, I have to implement the LDAP codec over MINA 3, which can be > a bit challenging, but could be a perfect move to test the migration > process, and see what's missing. > > > -- > Regards, > Cordialement, > Emmanuel Lécharny > www.iktek.com > > -- Jeff MAURY "Legacy code" often differs from its suggested alternative by actually working and scaling. - Bjarne Stroustrup http://www.jeffmaury.com http://riadiscuss.jeffmaury.com http://www.twitter.com/jeffmaury
