Re: Developing Applications using Mathematica
- To: mathgroup at smc.vnet.net
- Subject: [mg77122] Re: Developing Applications using Mathematica
- From: David Reiss <dbreiss at gmail.com>
- Date: Mon, 4 Jun 2007 03:51:15 -0400 (EDT)
- References: <f3r96l$t5f$1@smc.vnet.net>
Hi Mark, Since no one else has chimed in on this perhaps I will take a crack at it. I think though that some of what I will say involves a bit of conjecture and also is tingled by what I have been doing lately (http://scientificarts.com/worklife). However perhaps it can lead to an interesting discussion, and also perhaps attract some additional comments by the good folks at WRI. Your questions are particularly pertinent now that Mathematica 6 is out. The lines that were somewhat blurred previously are now blurred fully, and the expectation is that deployed applications with Mathematica will become quite commonplace as time moves on from the May 1st release of M 6. Of course I am particularly interested in this from both a personal and business perspective because of http://scientificarts.com/worklife which I intend to continue to develop--it's the main tool that I use to interact with Mathematica, and it's also the main tool I use to stay within Mathematica as much as is possible so that I can do my work the most efficiently (and also, I have to say, the most enjoyably). As for the question in your second paragraph--taking that first--I suspect that the long term plan for Mathematica Player is exactly what you are asking about. In the past WRI made use of a technology for creating Custom Mathematicas, and this was what they used for creating Calculation Center, NKS Explorer, Mathematica Teacher's Edition, and so on. I suspect that the plans for Mathematica Player--making use of the interface elements of Mathematica 6 (and in particular functions such as Deploy that allow one to **distribute** interfaces in a way that they cannot be modified other than the specified interface elements such as buttons, check boxes & c) --are to allow for the creation and deployment Mathematica applications that do not require the purchase of a full-blown Mathematica. Of course the initial examples of some of this that simply make use of Manipulate are the 1300 plus small applications that are on the Demonstrations site. So, in a word, I think that the answer to your second paragraph's question is "yes." After this, there are additional words and questions that have to do with when WRI will have the infrastructure (including legal License specifications) in place to anoint applications intended to be run under Mathematica Player for various potential types of distribution. Getting Mathematica 6 out was quite a feat, so I expect that having the full infrastructure for deployed applications based on Mathematica 6 and mathematica Player in place is one of the things that is intended for sometime in the (hopefully near) future. There are two questions, I think, in your first paragraph. One has to do with building applications in Mathematica directly and the other has to do with building applications that rely on Mathematica for their internal computations. Also perhaps a third is the use of Mathematica to prototype an application. One of the most obvious ways to use Mathematica is through webMathematica. I have created several such "enterprise" applications in the past for customers or employers. Usually these are to some fairly definite specification rather than as a general-use "tool box." Of course behind the scenes is a Mathematica package that provides the functionality, and in that sense it is a toolbox that can be expanded arbitrarily with a webMathematica deployment making use of it if an HTML rendition of a user interface can be created that suits its needs. webMathematica is quite powerful, and if the interface that you need for a deployable application is not too sophisticated and only needs the elemental HTML form entities, then quite powerful applications can be created and deployed--and they can quite easily be linked to a back- end database using Database Link if need be (which is what I did in one of the webMathematica apps that I created for a client). Another way, of course, is to create an application that requires a Mathematica Kernel and FrontEnd to run. All traditional add-ons are like this, and though I don't know WRI's pricing for network or site licenses, the cost per CPU process will certainly not be anything like the price of a single license for a individual Mathematica. And, if it is unlikely that more than some small fraction of the 1000 users will be using the application at any one time, then a network license with the number of processes optimized to the expected traffic would possibly make sense. Another possibility comes to mind (and again is one which I have used in the past). This is that Mathematica is an ideal environment in which to prototype the functionality of an application (as my daughter might say "well, duh!"). And, again, with the interface elements of Mathematica 6, this extends well beyond the internal algorithmics to a very arbitrary space of possible user interfaces. And I suspect that there will be considerable exploration of the space of possible user interfaces that will be catalyzed by Mathematica 6. (And this may well blossom as a field in itself.) This is both because the creation of user interfaces is **very** easy in Mathematica 6, and also because how one might think of user interfaces is greatly expanded from what one has been able to do in the past. After all, in Mathematica, user interfaces are expressions, as is everything else. And this is the most powerful, and new, aspect of it all. Exploring user interfaces--the space of user interfaces is indeed very NKS-like, which is certainly not surprising. To be fair, in Mathematica 5.x (and earlier) there were user interface elements that were quite powerful because of the fact that they are (and they are still there--Notebooks and buttons) themselves expressions. And in http://scientificarts.com/worklife I took these tools about as far as I could to expand the reach of Mathematica into areas that are helpful to folks. But the new elements in Mathematica 6 will allow me to take this **much** further. So this also gives a sense of my experience with creating "non- academic" applications. It is very doable, and the scale can be quite broad ((http://scientificarts.com/worklife has about 50,000 lines of Mathematica code). And I think that the option of deploying applications intended for broad deployment would be best done using Mathematica 6 and Mathematica Player. This would also --given the speed with which one can create applications in mathematica--make the turn-around time for modifications and enhancements very much quicker than with any other technology that I know of. It would be nice to hear from the gentle (and incredibly hard working) folks at WRI what the possible time line is for this approach. I am sure that discussions with the WRI partnerships group could make the possibilities for this clear enough so that you to make a decision on whether and how to bring this idea to your organization . Of course I'd be happy to help with this if you wish. (Consultants always have to say that...) ...hmmm, I seem to have gone on here a bit in a very bloggish way. I think that I will also place this note on http://scientificarts.com/worklife/wlfwblog/