On Mon, Mar 31, 2008 at 11:08:03AM +0200, Danny Chan wrote:

Hi everybody! I am currently looking into using squeak for data analysis at work. At the moment, I mostly need to postprocess data files created by other programs and put graphs and overview tables into some kind of PDF report. Are there any existing solutions for such tasks? Later, I would like to move some of the code for data analysis from older C programs to Smalltalk. Is there a capable matrix/vector library, like numpy/scipy for python?

PlotMorph is very useful (http://wiki.squeak.org/squeak/2626). In addition to providing graphs in Squeak, I found out almost completely by accident that the data points on a PlotMorph graph are "live". You can click on one, then explore the objects that they refer to. This is quite useful if you have a large amount of data with relationships that may not be obvious.

Dave

If you work in a command line/ batch job friendly environment, I can definitely recommend python with scipy and matplotlib. In my previous job I used this combination for simulation work on Unix platforms, and most of the times it was even faster than the dedicated Fortran programs written in the company.

I am looking into squeak because I am now in a company where Windows is used by most people (although Linux needs to be supported, too). Also it must be possible to use the tools on any computer without much setup work. My python scripts didn't work most of the time because of differences in the interpreter version installed, differences in the environment variables, sometimes even the different command line shells. Working with lots of scripts on the command line was ok on Unix, but it is a pain on Windows. Something like Smalltalk images would be handy here, and the development environment is really nice.

I think I will try to wrap a few open source libraries like plplot and rlib, maybe I can tackle larger problems once I have solutions for basic problems like generating a good looking graph.

Smalltalk really could be a great environment for scientists and engineers. The development tools are much better than what you get for Matlab for example. It just seems that the libraries are missing. But since Python has a lot of bindings to all kinds of scientific libraries (missing only the development environment), it should be possible to have something like this in Squeak/Smalltalk too.

It would be great to have something like ScientificSqueak just like Python has scipy...

Danny

Am Montag, 31. März 2008 17:18:43 schrieb David Finlayson:

Danny -

I am learning squeak for exactly the same reasons.

Up until a month or so ago, I did all my data analysis in C programs. A funny thing happened when my workstation was upgraded to an 8-core machine: my programs became I/O bound at a work load of 2 to 3 (ideal would be 8). My CPUs are way faster than my disk drives. This situation got me thinking about going back to dynamic languages since I've got plenty of cycles to burn on this new computer. Why not Python? Why not Smalltalk?

I have all the usual reservations about Squeak/Smalltalk. Does Squeak have the libraries to develop data analysis programs? Will it fit in to our existing infrastructure? Will people except and use software tools written in Squeak? Will I be able to share my code with colleagues? Not sure about the answers to any of these questions.

Right now I am just working my way through the books and tutorials to get a feel for the language and Squeak environment.

Cheers,

David _______________________________________________ Beginners mailing list Beginners@lists.squeakfoundation.org http://lists.squeakfoundation.org/mailman/listinfo/beginners