Hi Michael,

That's a very fair point.  Please know that Henri Gourvest and I have been
working together pretty closely on a number of aspects. I've very much
appreciated his help and support with a number of issues, and his
permission to incorporate some key functionality.

Speaking for my own project (
http://www3.nd.edu/~jeff/mathprog/mathprog.html ), the main goal is support
for teaching applications of OR methods. That's the reason behind keeping
the interface simple and incorporating an annotated list of examples.  My
secondary goal is more experimental -- how should an app be designed to
support MOOC scale teaching?  That's the reason for using Box, Dropbox, and
Google Drive for file stores, and maintaining the logic client-side for
integration with web based course management systems. (In fact, that's how
I expose MathProg to students -- as a web resource within Sakai, our local
course management system).

A web ide for applications development would probably make a different set
of choices. While there is substantial overlap, at a minimum you'd want a
more solver options, integration with external data feeds, and server side
support for project management. Maybe at some point a sufficiently polished
web ide for OR will be available that can span the range of use cases,
including teaching. In the meanwhile some diversity of thinking and
implementation may be a good thing.

One nice aspect of glpk.js is that, with the same code base, calculations
can be done client side in the browser or server side with Node.js. I could
be wrong on all of this -- wouldn't be the first time -- but I regard
Henri's development of glpk.js as a potential game changer for the wide
spread adoption of OR techniques. With little effort the same models can be
solved on phones, tablets, browsers, servers, distributed databases.  There
are tradeoffs, including performance penalties and frustrations with
sandboxed file stores. But that has to be balanced against opening glpk's
very nice api to web applications.  Consider the opportunities -- web based
route planning, portfolio management, scheduling, complex business logic,
smarter visualizations. There are some amazing js libraries out there to
support rapid development of complex applications.

Again, speaking for my own project, I don't think it should be seen as
anything more than addressing a specific use case of glpk.js --
introductory teaching of applications for ops research methods. In my own
view, glpk.js opens up many interesting opportunities to develop complete
IDE's and other great web apps, and hope others will join in the fun.

Jeff


On Fri, Feb 22, 2013 at 9:40 AM, Michael Nagel <[email protected]>wrote:

> Hi all,
>
> I have one more comment:
>
> with
> - http://www3.nd.edu/~jeff/mathprog/mathprog.html
> - http://mathp.org
>
> there seem to be two projects that offer quite similar functionality.
>
> I do not know how big the market is for Mathprog Web IDEs and what your
> ultimate goals are,
> but maybe it might be a good idea to join forces...
>
> Best Regards
> Michael
>
> _______________________________________________
> Help-glpk mailing list
> [email protected]
> https://lists.gnu.org/mailman/listinfo/help-glpk
>
_______________________________________________
Help-glpk mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/help-glpk

Reply via email to