toss-devel — Toss develpment discussion

Re: [Toss-devel] Rewrite of Mso.py

[Lukasz S. <luk...@gm...>]

My current take is to let the other side (e.g. sKizzo) sort out the
repetitions. Once I've got a benchmark, I'll write a dedicated CNF
structure without redundancy to compare if it's better to send less
bits across a channel.

On Fri, Aug 28, 2009 at 2:40 PM, Lukasz Stafiniak<luk...@gm...> wrote:
> OK, I know the reason.
>
> On Fri, Aug 28, 2009 at 1:09 PM, Lukasz Stafiniak<luk...@gm...> wrote:
>> Is there any reason behind extensive sorting when generating qdimacs?
>> I'm not going to do that.
>>
>

Re: [Toss-devel] Rewrite of Mso.py

[Lukasz S. <luk...@gm...>]

OK, I know the reason.

On Fri, Aug 28, 2009 at 1:09 PM, Lukasz Stafiniak<luk...@gm...> wrote:
> Is there any reason behind extensive sorting when generating qdimacs?
> I'm not going to do that.
>

[Toss-devel] Rewrite of Mso.py

[Lukasz S. <luk...@gm...>]

Is there any reason behind extensive sorting when generating qdimacs?
I'm not going to do that.

Re: [Toss-devel] Rewrite of Toss in OCaml

[Lukasz K. <luk...@gm...>]

> I will commit the files and fix the recent bug tonight. Perhaps I need
> to think a bit about the install suite (currently I'm using the
> Makefile, it's OCaml-specific as Toss hasn't been using "make".)

I think a Makefile is ok, as long as it is readable.
We are using a Makefile in the Client module,
because some part of it is in C for efficiency.

Lukasz Kaiser

[Toss-devel] Rewrite of Toss in OCaml

[Lukasz S. <luk...@gm...>]

Hi all,

I'm rewriting large parts of Toss into OCaml. Exactly how large is
open to debate. On the low end, at least the whole of "Mso.py" plus
some tools for representing models (parsing, printing). On the high
end, I could rewrite the whole of Toss eventually, including GUI...

"Mso.py" got already split into several files in the rewrite, mostly
because I'm using ocamlyacc for parsing.

My near-term interests are to write using Toss a GGP player, so I will
probably also add a GDL parser and some interaction protocol.

I will commit the files and fix the recent bug tonight. Perhaps I need
to think a bit about the install suite (currently I'm using the
Makefile, it's OCaml-specific as Toss hasn't been using "make".)

Cheers!

Re: [Toss-devel] Interpretation in the model

[Lukasz K. <luk...@gm...>]

>>> What's the status of an interpretation that is satisfied with an empty
>>> assignment, as in test_qbf_5? Isn't there supposed to be a guard that
>>> the assignment is nonempty?

You are right - test_qbf_5 is wrong, it should say that
the (only) element is true. I think the bug is in how
the singleton guards are generated for one element
structure - but whatever, there is a bug. It should say
x1 and (the rest) because interpretaions for translated
first-order variables can not be empty (we do not allow
empty structures - even if we do not check this yet).

Lukasz Kaiser

Re: [Toss-devel] Interpretation in the model

[Lukasz S. <luk...@gm...>]

No, I sustain the initial worry.

On Thu, Aug 27, 2009 at 2:05 PM, Lukasz Stafiniak<luk...@gm...> wrote:
> Silly me! Please don't explain :-)
> (I guess I need some sleep)
>
> On Thu, Aug 27, 2009 at 2:00 PM, Lukasz Stafiniak<luk...@gm...> wrote:
>> Hi,
>>
>> What's the status of an interpretation that is satisfied with an empty
>> assignment, as in test_qbf_5? Isn't there supposed to be a guard that
>> the assignment is nonempty?
>>
>> Best Regards,
>> Ł.S.
>>
>

Re: [Toss-devel] Interpretation in the model

[Lukasz S. <luk...@gm...>]

Silly me! Please don't explain :-)
(I guess I need some sleep)

On Thu, Aug 27, 2009 at 2:00 PM, Lukasz Stafiniak<luk...@gm...> wrote:
> Hi,
>
> What's the status of an interpretation that is satisfied with an empty
> assignment, as in test_qbf_5? Isn't there supposed to be a guard that
> the assignment is nonempty?
>
> Best Regards,
> Ł.S.
>

[Toss-devel] Interpretation in the model

[Lukasz S. <luk...@gm...>]

Hi,

What's the status of an interpretation that is satisfied with an empty
assignment, as in test_qbf_5? Isn't there supposed to be a guard that
the assignment is nonempty?

Best Regards,
Ł.S.

[Toss-devel] partial differential equations

[Michal R. W. <mic...@gm...>]

[Toss-devel] used nemerle snapshot

[Lukasz K. <luk...@gm...>]

Hi.

I recompiled speagram and toss using Nemerle from boot/
in snapshot from 11.01.2005 version 0.2.1.99.3731. To avoid 
confusion with multiple snapshots please use this one if possible.
In speagram there are still some warnings that have to be
corrected but except for this it compiles without problems.
The speagram archive inside toss is also updated to the
one compiled against the latest Nemerle snapshot.

- lk

[Toss-devel] Modal Dialogs within GTK

[Peter C. <pm...@gm...>]

Hi,

the issue I want to bring up is that is seems that GTK does _only_ allow 

modal dialogs when the parent window can be exposed to the dialog creating 

instance which then can set it as an appropriate field within its created 

dialog. 

Anyway the problem arises as to that it would require for a lot of functions
to carry this

field (parent window pointer) as a parameter in order to pass it as an
instance that

might want to create such a (modal) dialog.

Therefore, we might want to _really_ consider the option of having the GUI
create

the TsControls  (which would be realized over an interface that the plugin
manager sets

appropriately once loading a plugin and which over this interface can
request controls 

from the GUI). 

Furthermore, such a redesign would lead to a cleaner cut and dependability
of the controls.

I demand a discussion! ;-)

 

Greetings Piotr