[Clam-devel] Tests

Francisco Tufró nictuku at gmail.com
Wed Apr 9 07:10:44 PDT 2008


On Wed, Apr 9, 2008 at 10:24 AM, David García Garzón <dgarcia at iua.upf.edu>
wrote:

> El Tuesday 08 April 2008 22:33:44 Francisco Tufró va escriure:
> > Hello, i've started to do some tests related with the typed connection
> > templates. As soon as i have some result i'll post it to the list.
> > I ported the InControl class to a templated class.
> > Right now i'm hardcoding every instance of InControl as InControl<float>
> to
> > compile and test it.
>
> Before proceeding, i would suggest that you get the guidance of Pau or
> mine.
> This is a very transversal modification and it should be done carefully to
> keep the rest of the code compiling. We had hard experiences with
> transversal
> migrations such as this one. It needs some planning, test driven
> development
> and so on. Send to the list any patch/code you are working on so this
> could
> turn into a community effort.
>

You're right, i was just "playing" a bit to get in touch with the way it
works.
I've found some linking problems that i'm going to try to solve.
I've reused the name TControlData to refer the control data type, but
commented the "typedef float" and added the template stuff.
When talking about the code i found some problems.

1) In OutControl.hxx (and InControl.hxx the other way) you have the mLinks
member defined like this:

    std::list<InControl*> mLinks;

The problem is that if we have typed controls the elements from that list
can be of different types and this can't be handled with the std::list (i
think).
What do you think about that?

2) This is not a problem, The InControlTmpl class will be a 2 parameter
template.

template<class ProcObj, class TControlData>
class InControlTmpl : public InControl<TControlData>

That's all by now, if you want the code tell me, right now it doesn't link,
but i'll try to make it work.


>
> > Then i'll start doing some not float tests.
> > I've found that scons copies all the clam source into scons/lib/core/src
> > and asked hordia about it. He told me that's the way it works for clam.
> I
> > think this should be noticed somewhere, it's a strange behaviour, and
> the
> > errors appear with the scons path and not the real, that's a problem,
> one
> > has to find by hand (or by find -iname :P) the file.
>
> This is not that strange. Most projects move all the files to a distint
> folder
> such CLAM does. Anyway we don't like that, because errors are reported in
> files that you cannot edit. It is a long life to-do for the build system.
> I
> would rather invest time in fixing it than documenting it. ;-)
>
> David.
>
> that's ok :)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.clam-project.org/pipermail/clam-devel-clam-project.org/attachments/20080409/50a3736b/attachment.html>


More information about the clam-devel mailing list