Lsapi api

[jrmarino]

I seem to recall that the original intent for LSAPI is that it would be an alternative to the fastcgi protocol.

How is this effort going? Will LSAPI be documented and available for use for our own executable programs? It seem this has stalled and that's a shame, I'm sure people are interested in this.

[NiteWave]

lsapi has been there for years. and every release include them: for php and ror.

here's php lsapi document:
http://www.litespeedtech.com/php-litespeed-sapi.html

We did benchmark recently, here's the detail of lsapi vs fastcgi:

http://blog.litespeedtech.com/2010/0...ginx/#more-173

[jrmarino]

you misunderstood the question.

You publish PHP-LSAPI and Ruby-LSAPI. We all know that.

However, you guys said that you would publish the API for LSAPI so that we could use it for our own applications. E.g. for things other than PHP and Ruby. This was mentioned long ago as a goal. I see no movement in this goal.

Do you now understand what I am asking about?

[mistwang]

You can use our PHP-LSAPI and Ruby-LSAPI code as an example. All client side code is open, the API is pretty much like FCGI, we can provide a simple sample code like the "echo" FCGI example.

We do not spend much time to push it as an third party integration API as there is not much interest in it even though the code is completely open, pretty much all are internal development.

[jrmarino]

yeah, I knew I could dig around the php-lsapi code.

I think there "is not much interest" because it has never been published and documented. If somebody spent the time to properly document the API, then people would consider using it.

Given the benchmarks that nitewave just linked to, why would people not be interested in using it over fastcgi? Certainly if they are already litespeed clients, but perhaps promoting LSAPI for any application might result in new ones. It's a distinguishing feature that sets litespeed apart.

I personally think it would be worth litespeedstech's time to document it properly. I wouldn't think it would change much, so it should be a minimal effort to maintain once it was written. If it turns out nobody cares, it didn't cost much (a few hours time?), and it could potentially be very interesting to prospective clients.

[jrmarino]

By the way, had this already been published, maybe somebody would have created and shared the python and mono LSAPI interfaces by now. Per your own reference:
http://www.litespeedtech.com/support...ead.php?t=1883

That was 12 releases ago, right?
Document the interface, and maybe the community will come up with these and more, faster.