RE: Automatic code generation

[William Krekeler]

With regard to automatic code generation, I don't have any code but I have 
abstract thoughts (mostly questions). This would be really cool but why, ie 
what is the end goal or user? I ask these questions for two reasons. The first 
is that you need to know who the end user is before you define the syntax for 
the 'pseudo-code used as the input for automatic code generation. Is your 
syntax going to be more or less complex than the high-level scripting language 
and many functions that Octave already provides? If less complex will this 
hinder the end-users ability to generate useful code?

The second point, most of the people I have run into using octave/matlab are 
solving a specialized problem unique to their dataset. While an automatic code 
generator would likely ease their workload it is unlikely to be able to solve 
the whole problem. Result, the end-user is learning two languages, 1) the 
auto-code generation pseudo-code input syntax, and 2) octave. This violates 
KISS.

If you move forward, put just as much or more effort up-front into defining the 
input syntax than you actually do creating the underlying code that writes the 
code. I wish you the best of luck in accomplishing your goal.

William Krekeler