":" operator and extended categories

[victor]

Hello,

I am wondering if it is normal that the operator ":" is working for imported categories (in object) but not in categories extending categories declaring the predicate called with ":".

A little example :

:- category(one).
  pred.
:- end_category.

:- category(two, extends(one)).
  mypred :- :pred.
:- end_category.

The error I get is :

ERROR!    existence_error(procedure, pred)
          in clause: mypred:- :pred
          above line: 7

Maybe there is a good explanation, but I don't see why it shouldn't be a good idea.

[pmoura]

No problem in extending the functionality of the :/1 operator so that it can be used from within categories. The changes are trivial; I just need to perform some more tests before committing them.

[pmoura]

The extending functionality is now available starting with r4270. Please check your code with this new version if possible. Reopen this ticket if you found any regression when using the :/1 control construct.

[pmoura]

I've reverted the changes in r4270, removing the extended functionality that would have allowed the use of :/1 control construct from within categories that extend other categories. Despite the simplicity of the changes, this extended functionality would mess with the simple semantics of categories, which are based in a simple importing mechanism instead of an inheritance mechanism. Note that it is always possible to use the alias/3 predicate directive to call a predicate in an extended category that is overridden in the category containing the call.

[victor]

I am not sure to understand because I was not using ":" operator to call an overridden predicate, but to call a predicate from an extended category efficiently.

If I use the "::" operator to call a predicate declared in an extended category from another category, I have the feeling that it will look for it in all the inheritance tree of objects importing the category, and it will result in an inefficient call to the predicate (since I am sure that it is only declared in the category extended)

Am I wrong ? Is there another way to make this call efficiently ?

Sorry if it is the wrong place to ask :)

[pmoura]

Replying to victor:

Sorry if it is the wrong place to ask :)

Could you post your question on the discussion forums instead? Thanks.