+= does work for variables, even Strings.

AFAIR, there already has been a discussion about += and friends for properties.

In fact I don't see why the compiler not just generates "x.a := x.a + y" from "x.a += y" before doing any code generating, disregarding what x.a is. Of course an error message will occur if x.a is not writable for any reason.

Thanks,
-Michael

Blindly replacing "a += b" with "a := a + b" may lead to a semantic change if "a" is an expression with side-effects.
Nevertheless, I do think that implementing += for properties is a good idea, but unfortunately FPC developers disagree.

Alexander Klenin wrote

Blindly replacing "a += b" with "a := a + b" may lead to a semantic change if "a" is an expression with side-effects. Nevertheless, I do think that implementing += for properties is a good idea, but unfortunately FPC developers disagree.

I do see that also properties could show site effects when used in a += expression, as they are accessed twice. But is this (in nearly all cases) not exactly what the user would expect ?

-Michael

Most of the time, yes. But there are some subtleties with interfaces and reference counting, which could bite you in the leg.

(as they already have for some people).

Michael.

reference counting, which could bite you in the leg. (as they already have for some people).

I see. _Hidden_ side effects :)

-Michael

In our previous episode, said:

Moreover, that IMHO the c operators are mostly meant for porting of existing C code. And I never saw C code with properties :-)

This would result in the same error, because x.a is not an lval.

DoDi

Hans-Peter Diettrich wrote

This would result in the same error, because x.a is not an lval.

The example that make me ask here was Form1.Caption := Form1.Caption + '.';

-Michael