>
> Waldek,
>
> > The point is that real code computes sizes, variable lists
> > etc via functions and compiler should cope with that.
>
> Well, as I said, my approach is to design things in a way that does not
> involve functions. Unfortunately, one has to cope with the current
> Algebra library which is not perfect in this respect.
>
> Anyway, the usual approach would be to remove such potential
> dependencies and function parameters have to be constant in context. The
> latter can usually be achieved by introducing a function layer that
> get's all the type parameters as arguments.
The problem is that in interesting cases you get dependencies
between arguments. Then you have problem how to ensure that
needed relations hold. You may leave this unchecked (but
apparently you call such case a "bug"). So having smaller
number of independent parameters and computing the other
via functions seem better.
> Of course, that doesn't solve the problem that in some cases there is an
> inherent dependency between the arguments. My approach would be to
> improve the library and remove such dependencies whenever possible. But
> that is a huge task and already your an my opinion differs.
IME dependencies are essential and you limit yourself to toy
problems disallowing them.
> I think, I'm going to invest some more time in improving the test
> framework in FriCAS (before modifying Algebra). Only yesterday I learned
> how you seem to test regression via the .output files. That all should
> become a simple "make check" that returns a summary of which tests
> succeed and which failed.
I wrote few years ago about the 'norm-out' script. You give it
as argument path to previous build directory and path to current
build directory and it prints differences in .output files.
If build is done using different Lisps there is a lot of numerical
noise, but when using the same Lisp normal differences are quite
small. Anything more needs explanantion (usualy more differences
mean bugs, but sometimes may be due to deliberate change).
--
Waldek Hebisch
[email protected]
--
You received this message because you are subscribed to the Google Groups
"FriCAS - computer algebra system" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/fricas-devel?hl=en.
