[LogoForum] aUCBLogo's lexical scope

The message below is being cross-posted from the LogoForum. Please
reply here at comp.lang.logo and it will be cross-posted back to the
LogoForum. The original author of this message is
taobert@yahoo.nospam.fr.



Hello Logo World,

Just a question for Andreas (or for aUCBLogo users):

Why the lexical scoping of aUCBLogo behaves differently between
interactive mode and program mode ?

Here are examples:

--- Direct mode (copy of my aUCBLogo console)
; Welcome to aUCBLogo version 4.7

x = 1

be f
* output x
end

be g x
* output f
end

f
1* ;-)
g 0
0* ;-)* logically it should be 1

--- Program mode
be f
*** output x
end

be g x
*** output f
end

to test_scope
*** x = 1
*** print f
*** print g 0
end

1
1
;-)* results are correct

Regards,
Bertrand


__._,_.___
LogoForum messages are archived at:
http://groups.yahoo.com/group/LogoForum

The message below is being cross-posted from the LogoForum. Please
reply here at comp.lang.logo and it will be cross-posted back to the
LogoForum. The original author of this message is
Andreas.Micheler@gmx.nospam.de.


Hallo Bertrand!

> Why the lexical scoping of aUCBLogo behaves differently between
> interactive mode and program mode ?

Because it's a bug. I have not noticed it before, sorry.
I really should extend the /checks/ to cover more primitives.

> Here are examples:
>
> --- Direct mode (copy of my aUCBLogo console)
> ; Welcome to aUCBLogo version 4.7
>
> x = 1
>
> be f
> output x
> end
>
> be g x
> output f
> end
>
> f
> 1 ;-)
> g 0
> 0 ;-) logically it should be 1

Yes. But I can explain why it is misbehaving, see below. :-)

> --- Program mode
> be f
> output x
> end
>
> be g x
> output f
> end
>
> to test_scope
> x = 1
> print f
> print g 0
> end
>
> 1
> 1
> ;-) results are correct

The trouble is, in the parsing of the line "be g x",
if there's already a global variable ::x,
then it will be parsed as "be g ::x",
however if there's no global x,
it will be parsed as "be g g::x".
It *should* be parsed as "be g g::x" in any case, so it's a bug.
The bug is also in aUCBLogo-4.8-debug so far,
but I will correct it in the next debug build of 4.8.

I just had to add a few lines to Parse.cpp:
+ ObjectP olexufun=th->lexufun;
+ if (dynFlag)
+ th->lexufun=0;
+ else
+ th->lexufun=name->intern();

NodeP formals=parser_iterate(inln, inlimit, inhead, semi, _T('\n'),
name, inparen, col, tcol, lineStart, lines);
+ th->lexufun=olexufun;

But I still have to run the /checks/ to ensure it wont break something
else.

See also:
<http://www.physik.uni-augsburg.de/~micheler/aUCBLogo-4.7_scope_bug.png>
<http://www.physik.uni-augsburg.de/~m...cope_bug_corre
cted.png>

Thanks for the bug report!
Andreas

__._,_.___
LogoForum messages are archived at:
http://groups.yahoo.com/group/LogoForum




taobert@yahoo.nospam.fr wrote:

> The message below is being cross-posted from the LogoForum. Please
> reply here at comp.lang.logo and it will be cross-posted back to the
> LogoForum. The original author of this message is
> taobert@yahoo.nospam.fr.
>
>
>
> Hello Logo World,
>
> Just a question for Andreas (or for aUCBLogo users):
>
> Why the lexical scoping of aUCBLogo behaves differently between
> interactive mode and program mode ?
>
> Here are examples:
>
> --- Direct mode (copy of my aUCBLogo console)
> ; Welcome to aUCBLogo version 4.7
>
> x = 1
>
> be f
> * output x
> end
>
> be g x
> * output f
> end
>
> f
> 1* ;-)
> g 0
> 0* ;-)* logically it should be 1
>
> --- Program mode
> be f
> *** output x
> end
>
> be g x
> *** output f
> end
>
> to test_scope
> *** x = 1
> *** print f
> *** print g 0
> end
>
> 1
> 1
> ;-)* results are correct
>
> Regards,
> Bertrand
>
>
> __._,_.___
> LogoForum messages are archived at:
> http://groups.yahoo.com/group/LogoForum