K-user mailing list

[Robby Findler <robby AT eecs.northwestern.edu>]

Just to clarify from my side on this issue: I believe the model I've posted is buggy. It should say "strict(1)" in the and production in the grammar, not seqstrict. And, IIUC, K would respond exactly how I'd like it to respond when this is fixed.

The reason I posted this is that in the class I'm running, we came up with this model via suggestions I took from the class. I then expected the buggy model to behave a certain way, but I didn't see that behavior, so I wanted to confirm that my understanding of K was correct. Turns out I was missing something and I now understand what I'm missing and so I can explain this to the class and then we can fix the bug and continue on, building our IMP model (and our understanding of K).

Hope that clarifies.

Robby

On Wednesday, October 3, 2012, Patrick Meredith wrote:

You are definitely right, and don't worry, we often use a hole notation too.

My only thinking here is that we *do* want the ability to write a short circuiting and that will not evaluate that div/0.  The fact that this should be nondeterministic bothers me.  If this gets changed as Traian mentions, it seems that then we would need to write manual heating and cooling rules to make sure that and correctly short circuits.

-Patrick

On Wed, Oct 3, 2012 at 6:22 AM, Robby Findler <robby AT eecs.northwestern.edu> wrote:

What would this mean from the standpoint of the rules that seqstrict
generates? (I understand the seqstrict annotation as a shorthand for
generating rules, but I'm not sure that's right.)

I thought it would generate a rule something like

  A:KResult and B:BExp ~> everythingelse => A ~> (hole and B) ~> everythingelse

(apologies for my bad notation here "(hole and B)" is really something
with a freezer around it and some substitutions and I'm not sure how
to write "everythingelse" properly)

Such a rule should fire!

And indeed, if I just remove the 'and' shortcircuiting rule, then I do
see the heating rule fire.

Robby

On Tue, Oct 2, 2012 at 10:36 PM, Patrick Meredith <pmeredit AT gmail.com> wrote:
> Actually, that is a good point, and I'm not sure.  This is the behavior you
> would normally want, but I'm not sure why it is working.  My best guess is
> that seqstrict always plugs the heated value back in (cool) before it heats
> the second value, deterministically, and that is where it is reducing your
> short circuit (which doesn't care about the sort of the second argument).
> This is what we'd want, so my guess is that is how it works.  I haven't used
> seqstrict in any of my definitions.  Someone more knowledgeable can probably
> confirm if I am correct.
>
> -Patrick
>
> On Oct 2, 2012 10:27 PM, "Robby Findler" <robby AT eecs.northwestern.edu>
> wrote:
>>
>> But why should that preclude the heating/cooling from applying?
>>
>> Robby
>>
>> On Tue, Oct 2, 2012 at 10:26 PM, Patrick Meredith <pmeredit AT gmail.com>
>> wrote:
>> > One definite issue is bool is in your syntax, so false and _ can be
>> > applied
>> > before any heating or cooling ever occurs.
>> >
>> > -Patrick
>> >
>> > On Oct 2, 2012 10:20 PM, "Robby Findler" <robby AT eecs.northwestern.edu>
>> > wrote:
>> >>
>> >> Yes, but the left is already a KResult so the heating (or is it
>> >> cooling?) rule should still fire. No?
>> >>
>> >> Robby
>> >>
>> >> On Tue, Oct 2, 2012 at 10:16 PM, Patrick Meredith <pmeredit AT gmail.com>
>> >> wrote:
>> >> > Hi Robby,
>> >> >
>> >> > Replace seqstrict with strict.  Seqstrict is deterministic
>> >> > sequentially
>> >> > (left to right)
>> >> >
>> >> > -Patrick
>> >> >
>> >> > On Oct 2, 2012 7:12 PM, "Robby Findler" <robby AT eecs.northwestern.edu>
>> >> > wrote:
>> >> >>
>> >> >> Hi all: I'm having trouble understanding a K program's behavior and
>> >> >> I
>> >> >> wondered if I could get some help.
>> >> >>
>> >> >> Here's the program:
>> >> >>
>> >> >> module IMP
>> >> >>   syntax AExp ::= Int
>> >> >>                 | "(" AExp ")"    [bracket]
>> >> >>                 | AExp "/" AExp   [seqstrict, left]
>> >> >>   syntax BExp ::= Bool
>> >> >>                 | "(" BExp ")"    [bracket]
>> >> >>                 | AExp "<=" AExp  [seqstrict]
>> >> >>                 > BExp "and" BExp [seqstrict]
>> >> >>
>> >> >>   syntax KResult ::= Bool | Int
>> >> >>
>> >> >>   rule <k> I1:Int / 0 ...</k> => <k> "div0" </k>
>> >> >>   rule I1:Int <= I2:Int => I1 <=Int

--
-Patrick