According to Frank Heckenbach:

Does that mean that protected (not protected var) parameters are passed by value? Especially in the case of large structures, can this be inefficient, taking up stack space and consuming time to copy the arguments?

Yes. :-(

Since protected parameters are read-only anyway, actually I see no need to distinguish between passing them by value or by reference. Of course, it would be best if the compiler could automatically choose the more efficient way of passing (i.e. small structures by value, and large ones by reference), no matter if declared as "protected", "protected var" or "const".

I agree. But ...

Or does the standard prescribe something different?

Yes. It requires that the value of the protected formal parameter cannot change during the execution of the function, so things like the following require it to pass everything by value:

Var X: Something;

Procedure foo ( protected a: Something; Var b: Something );

begin (* foo *) b:= SomeValue; end (* foo *);

begin foo ( X, X ); end.

(One might ask what purpose `protected' serves when it requires to pass everything by value anyway ... ;-)

Also, defining "Const" as "protected Var" is not completely compatible to BP, since in BP you can pass a calculated expression, not only an "lvalue" as a const parameter.

I have planned to enable this, too. Then `Const' parameters will indeed be something different than `protected' ones. It could be a good idea to pass small things by value for `Const' parameters ... and relatively easy to implement ...

Okay, I am working on it, but ...

But of course, these are no urgent issues, as long as one form of protected/const parameters works at all... :-)

... with low priority. ;-)

Peter

Dipl.-Phys. Peter Gerwinski, Essen, Germany, free physicist and programmer peter.gerwinski@uni-essen.de - http://home.pages.de/~peter.gerwinski/ [970201] maintainer GNU Pascal - http://home.pages.de/~gnu-pascal/ [970125]