[fpc-pascal] Is this behavior related to calling convention?

[Xiangrong Fang]

I have this function:

function Compare(v1, v2: Integer): Integer;

And called like this:

Result := Compare(stack.Pop, stack.Pop);

The behavior is that the second Pop is evaluated first, thus reversed the
meaning of the function.

Is it true that all parameter evaluation goes from right to left?

Thanks!
Xiangrong
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal

Re: [fpc-pascal] Is this behavior related to calling convention?

[Jonas Maebe]

On 07 Dec 2013, at 09:40, Xiangrong Fang wrote:

> Is it true that all parameter evaluation goes from right to left?

The order in which parameters are evaluated is unrelated to the calling 
convention. This order has always been undefined in FPC, and it is also 
undefined in recent versions of Delphi.


Jonas___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal

[fpc-pascal] Re: Is this behavior related to calling convention?

[leledumbo]

Nope, calling convention only specifies parameter passing order, not
parameter evaluation order. If you want to specify exactly how the
parameters are evaluated, put each in a local variable first, according to
your order needs.



--
View this message in context: 
http://free-pascal-general.1045716.n5.nabble.com/Is-this-behavior-related-to-calling-convention-tp5717622p5717624.html
Sent from the Free Pascal - General mailing list archive at Nabble.com.
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal

Re: [fpc-pascal] Is this behavior related to calling convention?

[Jürgen Hestermann]

Am 2013-12-07 10:15, schrieb Jonas Maebe:
> The order in which parameters are evaluated is unrelated to the calling 
convention.
> This order has always been undefined in FPC, and it is also undefined in 
recent versions of Delphi.

I am wondering what the advantage is to *not* garantee a left to right order.
I can't think of a situation where changing this order makes any sense.

___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal

Re: [fpc-pascal] Is this behavior related to calling convention?

[Marco van de Voort]

In our previous episode, J?rgen Hestermann said:
> I am wondering what the advantage is to *not* garantee a left to right order.

Undefined order allows to do the evaluation that needs the most temps first I
guess, which reduces spilling.

___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal

Re: [fpc-pascal] Is this behavior related to calling convention?

[Jonas Maebe]

On 07 Dec 2013, at 18:50, Jürgen Hestermann wrote:

> I am wondering what the advantage is to *not* garantee a left to right order.

http://svn.freepascal.org/cgi-bin/viewvc.cgi/trunk/compiler/ncal.pas?view=markup#l3268


Jonas

___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal