[03:39] *** Uraeus has joined #kamaelia |
[04:58] *** salmon_ has joined #kamaelia |
[05:24] *** Lawouach has joined #kamaelia |
[06:37] *** MS- has joined #kamaelia |
[06:37] < MS-> greetings |
[10:50] *** ronny < 3 py.test -f |
[11:23] *** Uraeus has joined #kamaelia |
[11:55] *** Uraeus has joined #kamaelia |
[13:37] < MS-> Working on fixing JMB's http "fixes". Includes actually making the code work and reverting bad decisions on core code. |
[13:51] < MS-> Sometimes I wonder how anyone was willing to offer up code for merge which didn't actually work |
[13:52] < MS-> Especially after spending an entire summer complaining about how broken other people's code is. |
[14:36] < MS-> Right. OK. Having gone through all the "HTTP Improvements", I've found that all that aside from minor tweaks and seemingly pointless renames of a few things, that there's either |
[14:36] < MS-> very few actual improvements |
[14:36] < MS-> or actually the improvements on the branch have already been merged |
[14:37] < MS-> which means they originated in my Scratch branch originally and weren't developed any further |
[14:37] < MS-> ie JMB didn't do the stuff I expected him to as part of his GSOC project (on the HTTP code this is) |
[14:38] < MS-> Indeed it potentially introduces a subtle bug into some code. |
[14:38] < MS-> I'm unclear whether it existed beforehand though, since the previous logic was even more convoluted. |
[14:39] < MS-> Over all the code is an improvment. I think. But needs a lot of work to be clean IMO |
[16:05] < MS-> anyway, since it's overall an improvement it's now merged. |
[16:05] < MS-> Overall increasing quality is IMO more realistic/practical. |
[17:11] < MS-> OK, now moving onto JMB's wsgi code |
[17:34] < MS-> I really like diff |
[17:34] < MS-> It really is my friend |
[17:34] < MS-> :) |
[18:19] < MS-> OK, I normally rag on the concept of pythonic vs non-pythonic, but this is just wierd |
[18:19] < MS-> in an __init__.py file there's: |
[18:19] < MS-> from _factory import WSGIFactory, SimpleWSGIFactory |
[18:20] < MS-> I don't understand that in this context at all |
[18:24] < eikenberry> What don't you get? |
[18:25] < eikenberry> Is it pulling (?)classes from a _private module into __init__? |
[18:26] < eikenberry> Ie. why have the _private module. |
[18:32] < MS-> It's just particularly convoluted in this context |
[18:33] < MS-> There's also Factory classes in this code |
[18:34] < MS-> which again don't make sense because they just create instances of a known class |
[18:34] < MS-> I'm all in favour of indirection where it makes sense, but in this case it just doesn't |
[18:35] < MS-> I'm looking at this FWIW: http://kamaelia.googlecode.com/svn/branches/private_JMB_MPS_WSGI/Kamaelia/Kamaelia/Protocol/HTTP/Handlers/WSGI/ |
[18:37] < MS-> I think I'm going shift all of that code over as an app, and then sort it out |
[18:37] < MS-> (I like code to have clean simple interfaces) |
[18:40] < eikenberry> Not much beats simple. |
[18:43] < MS-> I don't think many people looking to host a WSGI app using kamaelia would particularly care for this interface |
[18:43] < MS-> The actual handler itself is fine, but how it's configured is wierd :) |
[18:45] < MS-> You have to go through 2 layers of indirection and a factory method which is configured against an URL list which itself is then matched against a regex *before* picking the appropriate App |
[18:45] < MS-> to run |
[18:45] < MS-> Strikes me as very java-esque actually |
[18:45] < MS-> In java this would probably be considered pretty nice |
[18:46] < MS-> I think I'm certain what I'll do with this is switch it over to being a user specific app namespace, and then cherry pick into the main namespace. |
[18:53] < eikenberry> I agree... that does have a bit of a java feel. And having just reviewed a bunch of wsgi frameworks recently I don't find that very attractive. |
[18:54] < eikenberry> Could have worked on Zope/Plone as well... they have that same over-engineered feel. |
[19:23] *** MS- nods |
[21:15] *** Lawouach has joined #kamaelia |