[Gecode] FW: PowerB

Christian Schulte schulte at imit.kth.se
Tue Nov 11 11:08:12 CET 2003


> Question:
> You'll probably want these functions inlined (even though 
> they are not 
> called in search :-) ), so I have to figure out more about 
> your conding 
> conventions, is this about right?
> Declarations (class.hh) are collapsed by class families into 
> familyName.hh Implementations (class.cc) are collapsed for 
> all specail cases of a 
> propagator into propagatorName.cc
> Inline implementations (class.icc) go in propagatorName.hh

Correct. On top of that, there are two different inlines:

	inline:	the normal one, reads as: inline if the compiler thinks it
is worth it

	forceinline: I KNOW FOR SURE THAT THIS MUST BE INLINED. To the
compiler it means, do it or you'll be laid off. Unfortunately, gcc has no
means to tell that. MSVC has it, though. But even if the compiler is not
able to take advantage of that, it serves nicely as documentation.

>
> Please do.

Okay.

> >Computer scientist use base 2, of course. Reason: exact and fast 
> >algorithms.
> >
> >  
> >
> How do you squeeze it out of mlib?

Do you mean, libm? You can't, you have to do it yourself, which ought to be
easy.

> I've never used automatic test case derivation, but it sure 
> sounds cool :-)

It's not just cool, this is what has to be done as manual testing is not
really providing sufficient coverage... But as said, I have no real idea
myself.

Cheers
Christian




More information about the gecode-users mailing list