Am Mo, 18.09.2006, 12:14, schrieb roucaries bastien:
>> > and in the second time allow
>> > to create qucs wizard in python or octave for instance.
>> I don't get yet the point. Please explain what you would like to
>> in detail.
> 1. Call python and octave script from octave or python
For this you don't need a change in qucs. You can just do it. Did you
mean: "Call python and octave script from qucs"?
> 2. Call qucs function from python or octave
> Both way are useful for my purpose
>> > Personnally I have a lot of octave script about filter conception, and
>> > i will like to call them from qucs. A lot of script use an original
>> > method develloped in my uni called "equivalent schematic", that are in
>> > fact a graphical but nethertheless exact method to solve modal
>> > equation in guided structure.
>> That is for filter synthesis right?
>> And you coded it in Octave?
>> Now you want it have run from within the Qucs filter synthesis program.
>> Did I understand that correctly?
> Sure, and sorry for my bad english last day I was really tired and I
> made stupid mistakes. My code is mainly about wave guide filter.
Ok. I try to summarize:
* you have some filter synthesis code for waveguide filters written
* you want a GUI (a "wizard"?) for this code
* you think, Qucs could be a really nice GUI for that
So what you would like is:
* use the Qucs GUI to steer (run) your octave code
* and get the results back into the GUI
Is this correct as well?
>> BTW: Vincent already made some more efforts regarding filter code. He
>> added odd Cauer, zig-zag, and chebychev 2 (inverse) filters to his code
>> and we are right now figuring out how to use his code... Right,
>> So what will your code be about?
>> > I join a first attempt to create a base lib. Do you think it is worth
>> > the effort.
>> > If yes, that do you think about gsl (gnu scientific library).
>> > Personnally I believe that we could use a lot of gsl functions
>> > (particularly some integral functions) in qucs because:
>> > o this lib is well tested (used daily in cern)
>> > o I do not like to badly reinvent some numerical method.
>> The only thing I saw in this proposed lib was some numerial functions,
>> actually there seems no reason to use these if you intend to use GSL
> It is not only numerical function but also data type and arithmetic
> (vector, matrix, complex).
I didn't see that in the qucs-lib_0.0.11.orig.tar.gz you sent to us,
In fact there is a 'libqucsator.a', but it is not yet meant to be
re-used. Thus I had a thought already in the direction, but didn't
realize it as probably needed.
> We could alos create a lib for the equation parser, and therefore use
> it in the gui
> or in linecal dialog (for instance i will write 1/sqrt(2) instead of
> compute manually in another windows).
> The code joined is only for demonstration purpose, but i could expand
> if you agree about the purpose of the lib.
No need right now. I got the point about how the library should look
> About gsl it is another problem. In fact you duplicate a lot gsl (for
> example elliptical integral) and I ask if it will be better instead of
> coding ourselve these function to code only a c++ warper to gsl. I
> believe, we will gain time and correctness. Moreover gsl is availlable
> on a lot of platform and gpl. The downside is a dependency.
Right. I tried to avoid the dependency in the past.
>> I do not yet see the purpose of this library yet. Please explain.
> For instance, I need to use in wizard program, complex number and
> elliptical integral it will be nice to use the same function and data
> format than qucs-core.
> If I need matrix arithmetic, it will be nice to use the natice class of
See my above comments.
I am sorry, but did not yet understood fully the purpose of the efforts
of you so far. Still need some clarifications. Depending on the above
questions about your plan we can discuss what may be needed to get it
done and how it is done.
Sorry for my clumsy English as well,