Experience Report: Scheme in Commercial Web Application Development

Interesting report by Noel and his colleagues, that for some reason was announced on the PLT Scheme blog and not here...

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Modesty prevents...

I didn't think it was good form to post my own work on the homepage, but of course I'm glad someone else has!

Of course. You should have

Of course. You should have alerted me so that I could decide if to post a message about the paper. We don't encourage people to post about their own work...


On your PLT blog you note that the Flapjax component was not included in the paper -- is that available somewhere?

You might find the LtU

You might find the LtU discussion of Flapjax to be of interest.

LtU flapjax discussion

Thanks, but I'm more interested in Noel, et al's perspective of actually _using_ Flapjax in a more commercial environment, ie, the 'missing section' on the paper for this thread. If that fits better under the flapjax topic, I'm happy to have the conversation there.


Flapjax in action

A previous blog post describes our experiences with Flapjax. The section in the paper was more or less the same thing — I just wanted to highlight Flapjax again in case people were wondering why it wasn't in the paper.

My opinion is that Flapjax is REALLY good for handling complex events on a page. Our Flapjax code is about half the size of the equivalent pure Javascript and I find the programming scales better. With pure Javascript I have found the code complexity increases exponentially with the complexity of the event flow, whilst with Flapjax it feels linear (this is just my gut feel; I haven't quantified this in any way). Performance, however, does not scale so well. It is absolutely fine for most applications, but if you have hundreds or thousands of events (as we did) you will have problems.

speed, and more

Jacob Baskin has done a fair bit of work this summer on improving the performance of Flapjax. Some improvements have been checked in, and more will be over time. Jacob has completely rewritten our conference manager, Continue, in Flapjax, and this has exposed many interesting things, both problems and abstractions. Plus we already know more about how to optimize such programs (see the LtU discussion on that paper).

Anyone sufficiently curious and/or frustrated by these issues -- apply to do a master's or PhD at Brown! (-: