Maintainers & gevent 1.0 issues

[Sylvain Prat]

Hello,

I'm in the process of fixing some bugs I encounter when using IE9 and Gevent 1.0 (which was officially released some days ago). I've already proposed a patch to fix an issue I discovered with IE9 (See https://github.com/abourget/gevent-socketio/pull/170). I also noticed that all the transports except websocket and flashsocket are broken with gevent 1.0, at least for my application. Too bad, I need the threadpool introduced in gevent 1.0! I will try to fix the issues if I can.

However, I see that there are many other pending pull request and no recent activity on https://github.com/abourget/gevent-socketio. Are you missing maintainers? Can somebody check in the pending pull requests please?

Thanks,
Sylvain

[Alexandre Bourget]

Hey Sylvain,

I'm short on time for still a couple weeks. If you fix issues, I'll review and merge your code. Adding tests would greatly help. We wanted to make a release some time ago.. but we haven't managed to do so yet because of conflicting PRs.. and nobody refactored them yet.
As gevent releases its newer version, it would make sense to release a compatible version very soon.

I would be immensely grateful for all the efforts that you put into that.. and thanks for the IE fixes.

Alexandre

--
You received this message because you are subscribed to the Google Groups "gevent-socketio" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gevent-socket...@googlegroups.com.
To post to this group, send email to gevent-...@googlegroups.com.
Visit this group at http://groups.google.com/group/gevent-socketio.
For more options, visit https://groups.google.com/groups/opt_out.

[Sylvain Prat]

John Anderson have merged my changes and a bunch of other pull requests, many thanks!

I am in the process of fixing all the transports and consequently, I have other fixes to push. I have already fixed a problem in the json-polling transport that was not escaping the data properly and I'm currently polishing my htmlfile fixes. I have nearly all transports working except xhr-multipart that I cannot test. You will receive another pull request very soon.

However, I have some questions :
- Are the jstests executed by travis-ci? It would be nice!
- I can't test the xhr-multipart: it seems that its support has been removed from socket.io.js. How can I test this transport? (it seems broken too, just by looking at the code)
- I am not sure if I fixed the htmlfile transport properly. I dug into the WSGI handler since this transport needs to keep the (GET) socket open, as in the websocket transports, but behaves like the polling transports otherwise (i.e. unidirectional with get and post). However, I don't fully understand what's going on in the WSGI handler. At first, I was thinking that a WSGI middleware would suffice, except maybe for the flash policy server : you would intercept the socket.io requests and pass the other requests to the application unchanged. Why a WSGIHandler is used instead of a simple WSGI middleware? Because we need to access the server socket? Also, why do we need to pass the control to the application in the case of a socket.io request (where the application just call socketio_manage in the examples I have seen)? So that the users can spawn whatever background jobs they want to feed the socket? Why the application is run via a spawn? Also, why don't we use the WSGIHandler's result and process_result() facilities for writing the response? I think the recent problems with gevent 1.0 are just due to that fact (chunked responses were triggered inadvertently).

FYI, I found a documentation that is really helpful to understand the socket.io transports: http://showmetheco.de/articles/2011/8/socket-io-for-backend-developers.html

Thanks in advance for you help!