eiffel, Few newbie syntax questions

Few newbie syntax questions

Author	Message
Artur Biesiadowsk #1 / 4	Few newbie syntax questions I've just started to learn eiffel so excuse me if any of my questions are goofy. Until now I was mainly C++/java guy. 1) Is there anything similar to namespaces/packages in eiffel ? I can just prefix class names with package abbreviation (which seems to be case in few eiffel libraries I've seen), but it is not scalable and with longer prefixes can be tiring to type in. 2) Is there any way to force variable to be reference type if class is defined as expanded ? (Opposite of expanded keyword in variable definition). 3) Let's suppose that there is class A, class B inheriting from A, class X, class Y inheriting from X. Class Y has method onlyInY. Class A has method m(x:X). Class B redefines method m using covariance to m(y:Y) and calls onlyInY inside this method. Now somewhere in program we have variable a:A; , which is in fact instance of B (real type was lost somewhere due to parameter passing, storing in array or anything). Now I call a.m(x). B.m() should be used I think - it is redefined after all. But B.m() relies on features that appear only in Y. Something is broken. How eiffel resolves that ? Maybe B.m(y:Y) do not really redefine A.m(x:X) only adds another method which will be choosen at runtime depending on parameter class ? Or methods are alwyas called depending on static context (it would be very not-OO) ? I'm lost here. Artur
Fri, 05 Apr 2002 03:00:00 GMT

Patrick Schoenbac #2 / 4	Few newbie syntax questions On Mon, 18 Oct 1999 10:55:36 +0200, Quote: >1) Is there anything similar to namespaces/packages in eiffel ? I can >just prefix class names with package abbreviation (which seems to be >case in few eiffel libraries I've seen), but it is not scalable and with >longer prefixes can be tiring to type in. No. Eiffel intentionally uses the class as only organisatorical unit. Classes can be grouped into clusters though. In case of name clashes, most compilers offer a mechanism to rename classes at compile time. In case you are using ISE Eiffel, this can be done in the Ace files using the "adapt/rename" meachnism. Quote: >2) Is there any way to force variable to be reference type if class is >defined as expanded ? (Opposite of expanded keyword in variable >definition). Non-expanded typed are always reference types. Quote: >3) Let's suppose that there is class A, class B inheriting from A, class >X, class Y inheriting from X. Class Y has method onlyInY. Class A has >method m(x:X). Class B redefines method m using covariance to m(y:Y) and >calls onlyInY inside this method. >Now somewhere in program we have variable a:A; , which is in fact >instance of B (real type was lost somewhere due to parameter passing, >storing in array or anything). Now I call a.m(x). B.m() should be used I >think - it is redefined after all. But B.m() relies on features that >appear only in Y. Something is broken. How eiffel resolves that ? If I understand your description right, we have a so-called polymorphic catcall. There have been some efforts to detect such calls at compile time, but no compiler-enforcable rules have been found yet. In the chapter about typing in OOSC2, you find more information on this issue. Best regards, Patrick -- ---------------------------------------- Patrick Schoenbach Interactive Software Engineering, Inc. URL: http://www.eiffel.com
Fri, 05 Apr 2002 03:00:00 GMT

Artur Biesiadowsk #3 / 4	Few newbie syntax questions Thanks for other answer, but this one still bugs me. Quote: > >2) Is there any way to force variable to be reference type if class is > >defined as expanded ? (Opposite of expanded keyword in variable > >definition). > Non-expanded typed are always reference types. Yes, but I'm talking about class itself being declared as expanded, not local variable. Is it possible to force local variable to be reference to such class, or another CLASS_REF has to be created (as it is done with INTEGER and friends). BTW, is expanded status inherited ? What if one superclass is expanded and second isn't ? Artur
Sat, 06 Apr 2002 03:00:00 GMT

Patrick Schoenba #4 / 4	Few newbie syntax questions On Tue, 19 Oct 1999 17:08:29 +0200, Quote: >Thanks for other answer, but this one still bugs me. [...] Quote: >Yes, but I'm talking about class itself being declared as expanded, not >local variable. Is it possible to force local variable to be reference >to such class, or another CLASS_REF has to be created (as it is done >with INTEGER and friends). No, this is not possible. If you do not want all instances of class X to be an expanded type, you will have to use declarations like foo: expanded X and not declare class X as expanded. Quote: >BTW, is expanded status inherited ? What if one superclass is expanded >and second isn't ? No. The expanded status is not inherited. Best regards, Patrick -- ---------------------------------------- Patrick Schoenbach Interactive Software Engineering, Inc. URL: http://www.eiffel.com
Sat, 06 Apr 2002 03:00:00 GMT

Page 1 of 1

[ 4 post ]