Something I don't understand with ImageSegments

Mariano Martinez Peck wrote:
> Hi guys:   I have this little (piece) of test:
>
>
>     | segment external internal root |
>
>     external := 'mariano'.
>     internal := true -> external.
>     root := false -> internal.
>     internal := nil.
>
>    segment := ImageSegment new
>         copyFromRoots: (Array with: root with: external)
>         sizeHint: 5000
>         areUnique: true.
>
>     self assert: segment outPointers size = 2.
>
>
> And that is green.
>
> However, if I change 'mariano' to Object new, I have that segment
> outPointers size is 3 instead of 2.
>
> I really don't understand why. Does someone see something ?
>

What outPointers do you expect to be present?

With my limited understanding of ImageSegment theory I would expect
there to be no outPointers in this example, because there are no
references from objects in the ImageSegment to objects outside the
ImageSegment, except for the well-known objects true and false.

Regards,

-Martin


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

Mariano Martinez Peck wrote:
>
> Exactly. I would expect to have only false and true in the outPointers,
> but not also the Object new. In the case of 'mariano' it is correct and
> only get false and true in the outPointers, but if I put Object new
> instead of 'mariano', in outPointers I get, true, false and that object
> I created.
>
> I don't see the difference between creating a string and an object.

I tried your example. I see as outPointers true, false, and Object. Not
the object you created, but the class Object.

But I also fail to see why you'd get the class of one of the referenced
objects but not Association.

Regards,

-Martin

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

Hi all,

I've been catching up on ImageSegment news, and came across this old post
from January,
and got interested in discovering the mindset behind the outPointers array.
I re-ran
Mariano's original test code, with a small change:  turned 'Object new' into
TestOutPointers, a new
subclass of Object, with 2 inst vars and one method.  I'd be interested in
getting feedback on
whether my guesses (in quotes below) are in the ball park.  The results seem
to explain why
Association is not in outPointers - what is captured there in the case of
adding a class
is a lot like what you would need for filling in a class template.  If you
add an instance of a class,
you just need the name of the class.

Some comments I have gleaned from the ImageSegment class on outPointers :
This array contains:
       - OOPs of all objects outside the segment that are pointed to from
inside segment.
       - outPointers that are necessary to turn the bits in the file into objects.

Here's the code:

| segment external internal root |

    external := TestOutpointers.

    internal := true -> external.
    root := false -> internal.
    internal := nil.

   segment := ImageSegment new
        copyFromRoots: (Array with: root with: external)
        sizeHint: 5000
        areUnique: true.

 ^ (segment outPointers ) -> (segment arrayOfRoots )

====== result =================
{ false.                              "object outside the segment that is
pointed to from inside segment."
Object.                              "superclass of TestOutpointers"
a MethodDictionary(#tryList->a CompiledMethod(1725:
TestOutpointers>>tryList) )       "MethodDictionary"
#('iv1' 'iv2')                         "inst var names"
('testing' tryList)                  "method category"
nil.                                    "object outside the segment that is
pointed to from inside segment."
#TestOutpointers.               "symbol representing name of new class"
Smalltalk.                          "environment that TestOutpointers is
assigned to"
#'IST-outPointers'.              "category name for TestOutpointers"
true.                                  "object outside the segment that is
pointed to from inside segment."
Metaclass.                         "for initializing class variables and
instance creation messages particular to a class."
Object class.                      "class of TestOutpointers superclass"
ClassOrganizer.                  "need this in order to determine whether
TestOutpointers has been changed in the local env"
#'as yet unclassified'.          "??"
}

->
{false->true->TestOutpointers. TestOutpointers}



Any feedback would be very welcome...

Cheers,
Sheri

--
View this message in context: http://n4.nabble.com/Something-I-don-t-understand-with-ImageSegments-tp1296982p1560990.html
Sent from the Pharo Smalltalk mailing list archive at Nabble.com.

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project