Re: [Axiom-developer] Projects Suggestions

[daly]

>I have a few suggestions regarding two topics in the Axiom projects list.

>For the browser based frontend, it may be worthwhile to look into the
>Jupyter project. This is the frontend of the IPython project but without
>the python backend, so as to allow any language to use the notebook
>frontend. There is a demonstration server up showcasing this at tmpnb.org.
>This showcases the python, julia, and r backends.

Axiom has a working browser front end (at least in theory). The front
end allows you to input expressions and the results are returned in
a newly opened <div> section so it works like a "notebook". There
are several "todo" subpaths. 

First is to move the hyperdoc information. Some of the static pages
exist already. I've implemented some lisp code to support  dynamic 
page information (e.g. database query, ancestors). This is queued
behind an effort to redesign the databases (do we really need to
do disk access anymore?)

Second, I have experimented with Axiom websocket support. This will
allow Axiom to do interactive graphics in the web page (as well as
be an "equation server" on a port). Unfortunately, websockets and I
don't seem to be getting along. There is no such thing as a simple job.



>For the continuous building and integration, Travis may be a good bet. It
>is a continuous integration service that plugs into github repos. It is
>located at travis-ci.org.

Axiom builds take about 30 minutes, a little more with the regression
test suite execution. A full CATS test suite run can take 24 hours.
So running many parallel VMs doing builds will likely bury a server.
I'm not so sure how travis-ci or jenkins-ci would feel about using
so much horsepower. On the upside, my office server will keep me
nice and warm this winter.

I suppose I ought to look at how the real world CI systems work in
more detail so the Axiom CI is reasonably compatible. Sigh.

Tim