Re: WebSockets

[Nala Ginrut]

2014年4月2日 PM7:54于 "Neil Jerram" <address@hidden>写道：
>
> [Now guile-user only, since that seems appropriate]
>
>
> On 2014-04-02 11:08, Nala Ginrut wrote:
>>
>> On Wed, 2014-04-02 at 10:29 +0100, Neil Jerram wrote:
>>>
>>> Hi there,
>>>
>>> I'm interested in adding support for WebSockets
>>> (http://tools.ietf.org/html/rfc6455) to Guile's web modules.  Is anyone
>>> else interested in - or possibly already working on - that?
>>>
>>
>> I was planing to implement websocket in Artanis web-framework and
>> there's half-baked code. Then I realized it's better to integrate with
>> the server, but Artanis hasn't written its own server. So it's
>> appreciated if anyone can do the job. ;-)
>
>
> Wow, Artanis looks quite sophisticated and I'm wondering now if I should use it for my work instead of (web ...) directly.  I need to take a closer look.
>

I'm glad you like it!
But oh, shame on me!  So many people are interested in it but I haven't done a better database handling.
Anyway, you can build a site if you can play sql directly.
Although it works now, I haven't released it yet. Because I want to release the first version with better config and database module.
The latest branch is wip-sql-mapping. There will be some changes compare to master. :-)

> One immediate thing that I noticed: some of its API is different from (web ...) but not obviously better or at a different conceptual level.  For example, your route context seems conceptually equivalent to (web request).  Is this just because you first wrote Artanis before (web ...) was available?
>

Actually, route context is a Artanis specific local environment of each request.
It's used to store info of features in Artanis, and passing necessary info in Artanis environment only, like url-remapping, sql-mapping, cookies inner processing...all the parts unrelated to http standard.
The http requests are handled by (web ...), and it's just encapsulated in route context. So their function doesn't duplicate.

> Regards,
>       Neil
>