Dear All,

Needs a decent browser, JavaScript >2015

http://www.dogelog.ch/

Currently swallows errors silently. Everything written
in Prolog itself, read/1, consult/1, etc.. and then cross
compiled into JavaScript. Not sure whether it can already

compile itself. But it has a text field and can add the
clauses in the text field and execute the directives in the
text field, and it has write/1 and nl/0 into the HTML document.

More care for the good boy upcoming.

Have Fun!

Jan Burse, 24.05.2021 #StaySafe
http://www.jekejeke.ch/

To help understand Dogelog, we made a little Q&A.
Here are the results:

Q: What does it mean that Dogelog was produced
by a cross compiler?
A: Dogelog is not enteirely written in JavaScript,
a great part was written in Prolog and then cross
compiled into JavaScript.

Q: Can Dogelog be run without Android?
A: Yes, since the result after cross compilation
was entirely JavaScript, you can run it everywhere
where you find JavaScript, no Android required.

Q: Can Dogelog be run without internet connection?
A: Yes, the compilation has only generated client
code, there is no server code, so once the JavaScript
is loaded, you can run it without a connection.

Q: Can Dogelog be run without a Browser?
A: Yes, its also possible to run Dogelog from within
node.js for example, which is headless.

Q: Can Dogelog be run from within a smartphone?
A: Yes, if you find a capable JavaScript environment
you can also run it on a small device as a smartphone.

Q: Can a smartphone use be demonstrated?
A: Yes we have updated the link http://www.dogelog.ch/ so
that it does automatic zooming, by adding these lines:

<style>
body {
font-family: verdana, sans-serif;
font-size: 2vw;
line-height: 1.4em
}
</style>

Here is a screenshot, even an offline use, where the
internet connection was cut off after Dogelog was loaded:
https://gist.github.com/jburse/f6ad8de5f2a2de54c3efdbbe654e8013#gistcomment-3759547

Q: Can a tablet use be demonstrated?
A: Yes the same link http://www.dogelog.ch/ also works
for tablets. Strangely the same Chrome browser has better
rendering on a Huawei tablet than on Smasung smartphone.
But this is an issue of HTML/CSS and not of Dogelog:
https://gist.github.com/jburse/f6ad8de5f2a2de54c3efdbbe654e8013#gistcomment-3759548

Q; For further questions about Dogelog, where should I go?
A: Documentation about Dogelog will soon be out, tickets
with bug reports or feature requests can be opened on GitHub at:
https://github.com/jburse/jekejeke-samples/issues

Mostowski Collapse schrieb:
> Dear All,
>
> Needs a decent browser, JavaScript >2015
>
> http://www.dogelog.ch/
>
> Currently swallows errors silently. Everything written
> in Prolog itself, read/1, consult/1, etc.. and then cross
> compiled into JavaScript. Not sure whether it can already
>
> compile itself. But it has a text field and can add the
> clauses in the text field and execute the directives in the
> text field, and it has write/1 and nl/0 into the HTML document.
>
> More care for the good boy upcoming.
>
> Have Fun!
>
> Jan Burse, 24.05.2021 #StaySafe
> http://www.jekejeke.ch/
>

It can already do a little Unicode:

:- X = 'hel𝄞lo', write(X), nl.

:- X = 'hel𝄞lo', atom_codes(X, L), write(L), nl.

Try it

hel𝄞lo
[104, 101, 108, 119070, 108, 111]

But I have no clue how the hell I can tease out a
Unicode codepoint categorization from JavaScript?

Mostowski Collapse schrieb am Donnerstag, 27. Mai 2021 um 16:41:04 UTC+2:
> To help understand Dogelog, we made a little Q&A.
> Here are the results:
>
> Q: What does it mean that Dogelog was produced
> by a cross compiler?
> A: Dogelog is not enteirely written in JavaScript,
> a great part was written in Prolog and then cross
> compiled into JavaScript.
>
> Q: Can Dogelog be run without Android?
> A: Yes, since the result after cross compilation
> was entirely JavaScript, you can run it everywhere
> where you find JavaScript, no Android required.
>
> Q: Can Dogelog be run without internet connection?
> A: Yes, the compilation has only generated client
> code, there is no server code, so once the JavaScript
> is loaded, you can run it without a connection.
>
> Q: Can Dogelog be run without a Browser?
> A: Yes, its also possible to run Dogelog from within
> node.js for example, which is headless.
>
> Q: Can Dogelog be run from within a smartphone?
> A: Yes, if you find a capable JavaScript environment
> you can also run it on a small device as a smartphone.
>
> Q: Can a smartphone use be demonstrated?
> A: Yes we have updated the link http://www.dogelog.ch/ so
> that it does automatic zooming, by adding these lines:
>
> <style>
> body {
> font-family: verdana, sans-serif;
> font-size: 2vw;
> line-height: 1.4em
> }
> </style>
>
> Here is a screenshot, even an offline use, where the
> internet connection was cut off after Dogelog was loaded:
> https://gist.github.com/jburse/f6ad8de5f2a2de54c3efdbbe654e8013#gistcomment-3759547
>
> Q: Can a tablet use be demonstrated?
> A: Yes the same link http://www.dogelog.ch/ also works
> for tablets. Strangely the same Chrome browser has better
> rendering on a Huawei tablet than on Smasung smartphone.
> But this is an issue of HTML/CSS and not of Dogelog:
> https://gist.github.com/jburse/f6ad8de5f2a2de54c3efdbbe654e8013#gistcomment-3759548
>
> Q; For further questions about Dogelog, where should I go?
> A: Documentation about Dogelog will soon be out, tickets
> with bug reports or feature requests can be opened on GitHub at:
> https://github.com/jburse/jekejeke-samples/issues
>
> Mostowski Collapse schrieb:
> > Dear All,
> >
> > Needs a decent browser, JavaScript >2015
> >
> > http://www.dogelog.ch/
> >
> > Currently swallows errors silently. Everything written
> > in Prolog itself, read/1, consult/1, etc.. and then cross
> > compiled into JavaScript. Not sure whether it can already
> >
> > compile itself. But it has a text field and can add the
> > clauses in the text field and execute the directives in the
> > text field, and it has write/1 and nl/0 into the HTML document.
> >
> > More care for the good boy upcoming.
> >
> > Have Fun!
> >
> > Jan Burse, 24.05.2021 #StaySafe
> > http://www.jekejeke.ch/
> >

In Java Unicode categorization is piece of cake.
But in JavaScript, how is it done? Is this some:

You're Gonna Need a Bigger
https://www.youtube.com/watch?v=2I91DJZKRxs

LoL

Mostowski Collapse schrieb am Donnerstag, 27. Mai 2021 um 20:43:31 UTC+2:
> It can already do a little Unicode:
>
> :- X = 'hel𝄞lo', write(X), nl.
>
> :- X = 'hel𝄞lo', atom_codes(X, L), write(L), nl.
>
> Try it
>
> hel𝄞lo
> [104, 101, 108, 119070, 108, 111]
>
> But I have no clue how the hell I can tease out a
> Unicode codepoint categorization from JavaScript?
> Mostowski Collapse schrieb am Donnerstag, 27. Mai 2021 um 16:41:04 UTC+2:
> > To help understand Dogelog, we made a little Q&A.
> > Here are the results:
> >
> > Q: What does it mean that Dogelog was produced
> > by a cross compiler?
> > A: Dogelog is not enteirely written in JavaScript,
> > a great part was written in Prolog and then cross
> > compiled into JavaScript.
> >
> > Q: Can Dogelog be run without Android?
> > A: Yes, since the result after cross compilation
> > was entirely JavaScript, you can run it everywhere
> > where you find JavaScript, no Android required.
> >
> > Q: Can Dogelog be run without internet connection?
> > A: Yes, the compilation has only generated client
> > code, there is no server code, so once the JavaScript
> > is loaded, you can run it without a connection.
> >
> > Q: Can Dogelog be run without a Browser?
> > A: Yes, its also possible to run Dogelog from within
> > node.js for example, which is headless.
> >
> > Q: Can Dogelog be run from within a smartphone?
> > A: Yes, if you find a capable JavaScript environment
> > you can also run it on a small device as a smartphone.
> >
> > Q: Can a smartphone use be demonstrated?
> > A: Yes we have updated the link http://www.dogelog.ch/ so
> > that it does automatic zooming, by adding these lines:
> >
> > <style>
> > body {
> > font-family: verdana, sans-serif;
> > font-size: 2vw;
> > line-height: 1.4em
> > }
> > </style>
> >
> > Here is a screenshot, even an offline use, where the
> > internet connection was cut off after Dogelog was loaded:
> > https://gist.github.com/jburse/f6ad8de5f2a2de54c3efdbbe654e8013#gistcomment-3759547
> >
> > Q: Can a tablet use be demonstrated?
> > A: Yes the same link http://www.dogelog.ch/ also works
> > for tablets. Strangely the same Chrome browser has better
> > rendering on a Huawei tablet than on Smasung smartphone.
> > But this is an issue of HTML/CSS and not of Dogelog:
> > https://gist.github.com/jburse/f6ad8de5f2a2de54c3efdbbe654e8013#gistcomment-3759548
> >
> > Q; For further questions about Dogelog, where should I go?
> > A: Documentation about Dogelog will soon be out, tickets
> > with bug reports or feature requests can be opened on GitHub at:
> > https://github.com/jburse/jekejeke-samples/issues
> >
> > Mostowski Collapse schrieb:
> > > Dear All,
> > >
> > > Needs a decent browser, JavaScript >2015
> > >
> > > http://www.dogelog.ch/
> > >
> > > Currently swallows errors silently. Everything written
> > > in Prolog itself, read/1, consult/1, etc.. and then cross
> > > compiled into JavaScript. Not sure whether it can already
> > >
> > > compile itself. But it has a text field and can add the
> > > clauses in the text field and execute the directives in the
> > > text field, and it has write/1 and nl/0 into the HTML document.
> > >
> > > More care for the good boy upcoming.
> > >
> > > Have Fun!
> > >
> > > Jan Burse, 24.05.2021 #StaySafe
> > > http://www.jekejeke.ch/
> > >

Ha Ha, AI is going servers:

HumanE AI Net Virtual Lab - Unimore, 2021
https://www.youtube.com/watch?v=KvcK5V48jx0

Whats the result, some payed service?
Dogelog will be client, and payed through

dogecoins of course. Just joking.

Dogelog is still a pubby. It already chockes on a 5x5
example of Numberlink. It doesn't chocke because of
garbage collection, it chockes because of a couple of

cardinal sins in its design. But the 3x3 example works.

Brute Force Numberlink Solver
https://gist.github.com/jburse/8c387824f82c239cbf02f3ba893bcf02#file-path-pl

Brute Force Numberlink Solver
Dogelog Version, Smaller Example
https://gist.github.com/jburse/8c387824f82c239cbf02f3ba893bcf02#file-path2-pl

The Dogelog version avoids (\+)/1, since this is not
yet delivered. To make Dogelog an itch more performant,
there are a dozen ideas in the air, a) adopting some

approaches from Jekejeke, and b) extending the instruction
set to more from the Albuferia compiler. So the pubby will
be a real bastard when grown up, a

Jekejeke and Albuferia mongrel.

Corr.: Typo
Dogelog is still a puppy.

Mostowski Collapse schrieb am Donnerstag, 10. Juni 2021 um 12:53:30 UTC+2:
> Dogelog is still a pubby. It already chockes on a 5x5
> example of Numberlink. It doesn't chocke because of
> garbage collection, it chockes because of a couple of
>
> cardinal sins in its design. But the 3x3 example works.
>
> Brute Force Numberlink Solver
> https://gist.github.com/jburse/8c387824f82c239cbf02f3ba893bcf02#file-path-pl
>
> Brute Force Numberlink Solver
> Dogelog Version, Smaller Example
> https://gist.github.com/jburse/8c387824f82c239cbf02f3ba893bcf02#file-path2-pl
>
> The Dogelog version avoids (\+)/1, since this is not
> yet delivered. To make Dogelog an itch more performant,
> there are a dozen ideas in the air, a) adopting some
>
> approaches from Jekejeke, and b) extending the instruction
> set to more from the Albuferia compiler. So the pubby will
> be a real bastard when grown up, a
>
> Jekejeke and Albuferia mongrel.

Dogelog is a further evolution of SWI-Prolog qsave. Arbitrary
directives can be also included in cross compiled code. See the
user manual of Dogelog, which came out just yesterday,

section "embedding". Its on GitHub:
https://github.com/jburse/dogelog-moon/blob/main/manual/package.pdf

This means Dogelog doesn't need cludges like initialization/1.
Subsequently examples like these work fine. Namely one can
try the following in the browser:

:- assertz(foo(bar)).
:- assertz(foo(baz)).

test :-
foo(X),
write(X), nl, fail.
test.

:- test.

https://twitter.com/janburse/status/1404056443876253703

The browser doesn't use batch cross compilation, but online
compilation. But it adheres also to the idea that a Prolog text
is a sequence of Prolog clauses and Prolog directives.

I tried the same in TauProlog, it didn't work sensibly. There is
some speculation on twitter, that it ignores the directive.
May I suggest to throw some error?

Silently swallowing some directives is not a good idea. What
if I used an existing directive, and had a typo in it? Like
:- dynamik(foo/1) instead of :- dynamic(foo/1) ?

Mostowski Collapse schrieb am Sonntag, 13. Juni 2021 um 19:29:53 UTC+2:
> Dogelog is a further evolution of SWI-Prolog qsave. Arbitrary
> directives can be also included in cross compiled code. See the
> user manual of Dogelog, which came out just yesterday,
>
> section "embedding". Its on GitHub:
> https://github.com/jburse/dogelog-moon/blob/main/manual/package.pdf
>
> This means Dogelog doesn't need cludges like initialization/1.
> Subsequently examples like these work fine. Namely one can
> try the following in the browser:
>
> :- assertz(foo(bar)).
> :- assertz(foo(baz)).
>
> test :-
> foo(X),
> write(X), nl, fail.
> test.
>
> :- test.
>
> https://twitter.com/janburse/status/1404056443876253703
>
> The browser doesn't use batch cross compilation, but online
> compilation. But it adheres also to the idea that a Prolog text
> is a sequence of Prolog clauses and Prolog directives.

I wrote in SWI-Prolog discourse:
In Dogelog I have written the tokenizer in Prolog,
using the two-level DCG approach.

Then some body responded:
There’s also Richard O’Keefes tokeniser (section 10.7 of The Craft of Prolog ).
It’s probably online somewhere, but I didn’t see it at Software by
Richard A. O’Keefe

Then I responded:
Yes I know. The current Dogelog, read, get_token and write is
now open source. I uploaded it a few days ago. It has already prefix,
infix and postfix operators:

Dogelog Loader
https://github.com/jburse/dogelog-moon/blob/main/devel/transpiler/loader.p

The loader is transpiled into JavaScript. Its not yet complete,
for example line comments and block comments are not yet
realized. And the character classification is only for ASCII range.

But should also run with SWI-Prolog.

I am currently thinking of making it more easier to run from within
a Prolog system, so as to prototype it easier. Currently my own
Prolog system throws a few warnings(*). Some of the warning

could be possibly turned off by a better style check Prolog flag.

(*) For example when overriding existing read/1.

> The loader is transpiled into JavaScript.

Actually the above statement could be misleading. It is transpiled
into the instructions of a Prolog VM. But these instructions are
represented as JavaScript arrays and JavaScript objects.

Somehow an alternative to the qsave approach. The Prolog VM
accepts Prolog clauses and arbitrary goals as Prolog directives.
No need of initialization/1 directive, you get the input order

execution that you write down. This is more difficult for
qsave, which departs from a memory image and not an
existing Prolog text.

Mostowski Collapse schrieb am Dienstag, 15. Juni 2021 um 16:24:54 UTC+2:
> I wrote in SWI-Prolog discourse:
> In Dogelog I have written the tokenizer in Prolog,
> using the two-level DCG approach.
>
> Then some body responded:
> There’s also Richard O’Keefes tokeniser (section 10.7 of The Craft of Prolog ).
> It’s probably online somewhere, but I didn’t see it at Software by
> Richard A. O’Keefe
>
> Then I responded:
> Yes I know. The current Dogelog, read, get_token and write is
> now open source. I uploaded it a few days ago. It has already prefix,
> infix and postfix operators:
>
> Dogelog Loader
> https://github.com/jburse/dogelog-moon/blob/main/devel/transpiler/loader.p
>
> The loader is transpiled into JavaScript. Its not yet complete,
> for example line comments and block comments are not yet
> realized. And the character classification is only for ASCII range.
>
> But should also run with SWI-Prolog.
>
> I am currently thinking of making it more easier to run from within
> a Prolog system, so as to prototype it easier. Currently my own
> Prolog system throws a few warnings(*). Some of the warning
>
> could be possibly turned off by a better style check Prolog flag.
>
> (*) For example when overriding existing read/1.

Couldn’t help it otherwise, its some old
school Niklaus Wirth bootstrapping technique.

Niklaus Wirth reported that he wrote the first Pascal compiler in Fortran.
https://wiki2.org/en/Bootstrapping_%28compilers%29

Mostowski Collapse schrieb am Dienstag, 15. Juni 2021 um 16:38:14 UTC+2:
> > The loader is transpiled into JavaScript.
> Actually the above statement could be misleading. It is transpiled
> into the instructions of a Prolog VM. But these instructions are
> represented as JavaScript arrays and JavaScript objects.
>
> Somehow an alternative to the qsave approach. The Prolog VM
> accepts Prolog clauses and arbitrary goals as Prolog directives.
> No need of initialization/1 directive, you get the input order
>
> execution that you write down. This is more difficult for
> qsave, which departs from a memory image and not an
> existing Prolog text.
> Mostowski Collapse schrieb am Dienstag, 15. Juni 2021 um 16:24:54 UTC+2:
> > I wrote in SWI-Prolog discourse:
> > In Dogelog I have written the tokenizer in Prolog,
> > using the two-level DCG approach.
> >
> > Then some body responded:
> > There’s also Richard O’Keefes tokeniser (section 10.7 of The Craft of Prolog ).
> > It’s probably online somewhere, but I didn’t see it at Software by
> > Richard A. O’Keefe
> >
> > Then I responded:
> > Yes I know. The current Dogelog, read, get_token and write is
> > now open source. I uploaded it a few days ago. It has already prefix,
> > infix and postfix operators:
> >
> > Dogelog Loader
> > https://github.com/jburse/dogelog-moon/blob/main/devel/transpiler/loader.p
> >
> > The loader is transpiled into JavaScript. Its not yet complete,
> > for example line comments and block comments are not yet
> > realized. And the character classification is only for ASCII range.
> >
> > But should also run with SWI-Prolog.
> >
> > I am currently thinking of making it more easier to run from within
> > a Prolog system, so as to prototype it easier. Currently my own
> > Prolog system throws a few warnings(*). Some of the warning
> >
> > could be possibly turned off by a better style check Prolog flag.
> >
> > (*) For example when overriding existing read/1.

Even the simpler same language case.

Mostowski Collapse schrieb am Dienstag, 15. Juni 2021 um 16:45:43 UTC+2:
> Couldn’t help it otherwise, its some old
> school Niklaus Wirth bootstrapping technique.
>
> Niklaus Wirth reported that he wrote the first Pascal compiler in Fortran..
> https://wiki2.org/en/Bootstrapping_%28compilers%29
> Mostowski Collapse schrieb am Dienstag, 15. Juni 2021 um 16:38:14 UTC+2:
> > > The loader is transpiled into JavaScript.
> > Actually the above statement could be misleading. It is transpiled
> > into the instructions of a Prolog VM. But these instructions are
> > represented as JavaScript arrays and JavaScript objects.
> >
> > Somehow an alternative to the qsave approach. The Prolog VM
> > accepts Prolog clauses and arbitrary goals as Prolog directives.
> > No need of initialization/1 directive, you get the input order
> >
> > execution that you write down. This is more difficult for
> > qsave, which departs from a memory image and not an
> > existing Prolog text.
> > Mostowski Collapse schrieb am Dienstag, 15. Juni 2021 um 16:24:54 UTC+2:
> > > I wrote in SWI-Prolog discourse:
> > > In Dogelog I have written the tokenizer in Prolog,
> > > using the two-level DCG approach.
> > >
> > > Then some body responded:
> > > There’s also Richard O’Keefes tokeniser (section 10.7 of The Craft of Prolog ).
> > > It’s probably online somewhere, but I didn’t see it at Software by
> > > Richard A. O’Keefe
> > >
> > > Then I responded:
> > > Yes I know. The current Dogelog, read, get_token and write is
> > > now open source. I uploaded it a few days ago. It has already prefix,
> > > infix and postfix operators:
> > >
> > > Dogelog Loader
> > > https://github.com/jburse/dogelog-moon/blob/main/devel/transpiler/loader.p
> > >
> > > The loader is transpiled into JavaScript. Its not yet complete,
> > > for example line comments and block comments are not yet
> > > realized. And the character classification is only for ASCII range.
> > >
> > > But should also run with SWI-Prolog.
> > >
> > > I am currently thinking of making it more easier to run from within
> > > a Prolog system, so as to prototype it easier. Currently my own
> > > Prolog system throws a few warnings(*). Some of the warning
> > >
> > > could be possibly turned off by a better style check Prolog flag.
> > >
> > > (*) For example when overriding existing read/1.

We decided to provide the following JavaScript calls:

init(): (JavaScript)
Initializes the Prolog system.
set_output(C): (JavaScript)
Sets the current output callback.
set_stage(S): (JavaScript)
Sets the current stage.
clear(): (JavaScript)
Resets the Prolog database of the current stage.
consult(T): (JavaScript)
Consults the given text into the current stage.
show(T): (JavaScript)
Writes the Prolog term canonical.

The JavaScript calls init() and consult() are rather self-explanatory.
But one has to be aware that consult() currently does not take some
sort of a file name, but rather a string constituting a Prolog text.

The JavaScript calls clear() and set_output() can be used to
redo and redirect a consult(). What is possibly a curriosity is the
JavaScript call set_stage(). For more information see:

Preview: Dogelog runtime smart Prolog notebooks. (Jekejeke)
https://gist.github.com/jburse/8c387824f82c239cbf02f3ba893bcf02#gistcomment-3781233

Preview: Dogelog can be embedded in stackoverflow. (Jekejeke)
https://stackoverflow.com/a/67990574/502187

Mostowski Collapse schrieb am Dienstag, 15. Juni 2021 um 18:09:45 UTC+2:
> We decided to provide the following JavaScript calls:
>
> init(): (JavaScript)
> Initializes the Prolog system.
> set_output(C): (JavaScript)
> Sets the current output callback.
> set_stage(S): (JavaScript)
> Sets the current stage.
> clear(): (JavaScript)
> Resets the Prolog database of the current stage.
> consult(T): (JavaScript)
> Consults the given text into the current stage.
> show(T): (JavaScript)
> Writes the Prolog term canonical.
>
> The JavaScript calls init() and consult() are rather self-explanatory.
> But one has to be aware that consult() currently does not take some
> sort of a file name, but rather a string constituting a Prolog text.
>
> The JavaScript calls clear() and set_output() can be used to
> redo and redirect a consult(). What is possibly a curriosity is the
> JavaScript call set_stage(). For more information see:
>
> Preview: Dogelog runtime smart Prolog notebooks. (Jekejeke)
> https://gist.github.com/jburse/8c387824f82c239cbf02f3ba893bcf02#gistcomment-3781233

Even a reference manual is now out:
https://github.com/jburse/dogelog-moon/blob/main/manual/package.pdf

Mostowski Collapse schrieb am Donnerstag, 17. Juni 2021 um 03:22:41 UTC+2:
> Preview: Dogelog can be embedded in stackoverflow. (Jekejeke)
> https://stackoverflow.com/a/67990574/502187
> Mostowski Collapse schrieb am Dienstag, 15. Juni 2021 um 18:09:45 UTC+2:
> > We decided to provide the following JavaScript calls:
> >
> > init(): (JavaScript)
> > Initializes the Prolog system.
> > set_output(C): (JavaScript)
> > Sets the current output callback.
> > set_stage(S): (JavaScript)
> > Sets the current stage.
> > clear(): (JavaScript)
> > Resets the Prolog database of the current stage.
> > consult(T): (JavaScript)
> > Consults the given text into the current stage.
> > show(T): (JavaScript)
> > Writes the Prolog term canonical.
> >
> > The JavaScript calls init() and consult() are rather self-explanatory.
> > But one has to be aware that consult() currently does not take some
> > sort of a file name, but rather a string constituting a Prolog text.
> >
> > The JavaScript calls clear() and set_output() can be used to
> > redo and redirect a consult(). What is possibly a curriosity is the
> > JavaScript call set_stage(). For more information see:
> >
> > Preview: Dogelog runtime smart Prolog notebooks. (Jekejeke)
> > https://gist.github.com/jburse/8c387824f82c239cbf02f3ba893bcf02#gistcomment-3781233

Why didn't O-Prolog have correct if-then-else? Don't
blame ISO. Currently we are working our way through
disjunction (;)/2 support for Dogelog runtime:

Preview: Dogelog runtime can do Prolog disjunction. (Jekejeke)
https://gist.github.com/jburse/76c67a1529f306c9d376b3264a6a52b3#gistcomment-3784948

Whats also planned support for if-then-else (->)/2. And
a new compilation scheme for '$ALT'/1, which we
didn't need in Jekejeke Prolog.

Mostowski Collapse schrieb am Donnerstag, 17. Juni 2021 um 10:46:35 UTC+2:
> Even a reference manual is now out:
> https://github.com/jburse/dogelog-moon/blob/main/manual/package.pdf
> Mostowski Collapse schrieb am Donnerstag, 17. Juni 2021 um 03:22:41 UTC+2:
> > Preview: Dogelog can be embedded in stackoverflow. (Jekejeke)
> > https://stackoverflow.com/a/67990574/502187
> > Mostowski Collapse schrieb am Dienstag, 15. Juni 2021 um 18:09:45 UTC+2:
> > > We decided to provide the following JavaScript calls:
> > >
> > > init(): (JavaScript)
> > > Initializes the Prolog system.
> > > set_output(C): (JavaScript)
> > > Sets the current output callback.
> > > set_stage(S): (JavaScript)
> > > Sets the current stage.
> > > clear(): (JavaScript)
> > > Resets the Prolog database of the current stage.
> > > consult(T): (JavaScript)
> > > Consults the given text into the current stage.
> > > show(T): (JavaScript)
> > > Writes the Prolog term canonical.
> > >
> > > The JavaScript calls init() and consult() are rather self-explanatory.
> > > But one has to be aware that consult() currently does not take some
> > > sort of a file name, but rather a string constituting a Prolog text.
> > >
> > > The JavaScript calls clear() and set_output() can be used to
> > > redo and redirect a consult(). What is possibly a curriosity is the
> > > JavaScript call set_stage(). For more information see:
> > >
> > > Preview: Dogelog runtime smart Prolog notebooks. (Jekejeke)
> > > https://gist.github.com/jburse/8c387824f82c239cbf02f3ba893bcf02#gistcomment-3781233

Now I have the feeling I can inline anything, even stuff
that has cut or whatever. Cut (!)/0 inside call/1 is now
translated as follows:

?- G = (p(X), q(Y)),
transform_body(G, nothing, M, R, []),
transform_redo(M, R, S).
S = [p(X), q(Y)]

?- G = (p(X), !, q(Y)),
transform_body(G, nothing, M, R, []),
transform_redo(M, R, S).
S = ['$REDO'(_A), p(X), '$CUT'(_A), q(Y)]

Introduced the new internal built-in '$REDO'/1 yesterday,
together with '$SEQ/1 and '$ALT'/1. I guess it makes the
cut inlinable, just inline the transformed stuff.

This would mean inlining doesn't happen on the level
of term/goal expansion, but a little lower. But before
the encode, or maybe after the encoding?

P.S.: I borrowed '$SEQ/1 and '$ALT'/1 from Jekejeke Prolog
somehow, but the '$REDO'/1 and '$CUT'/1 stuff is
a new old idea, specifically pursued in the Dogelog runtime,

might later transfer back to Jekejeke Prolog, if successful
in all respects. The last proof of concept will be providing
if-then-else support based on '$REDO'/1 and '$CUT'/1.

Mostowski Collapse schrieb am Freitag, 18. Juni 2021 um 13:08:06 UTC+2:
> Why didn't O-Prolog have correct if-then-else? Don't
> blame ISO. Currently we are working our way through
> disjunction (;)/2 support for Dogelog runtime:
>
> Preview: Dogelog runtime can do Prolog disjunction. (Jekejeke)
> https://gist.github.com/jburse/76c67a1529f306c9d376b3264a6a52b3#gistcomment-3784948
>
> Whats also planned support for if-then-else (->)/2. And
> a new compilation scheme for '$ALT'/1, which we
> didn't need in Jekejeke Prolog.

Since inlining works, Dogelog is now old enough for its
own group. Dogelog got its own Facebook Group!

https://www.facebook.com/groups/dogelog

Mostowski Collapse schrieb am Freitag, 18. Juni 2021 um 13:58:21 UTC+2:
> Now I have the feeling I can inline anything, even stuff
> that has cut or whatever. Cut (!)/0 inside call/1 is now
> translated as follows:
>
> ?- G = (p(X), q(Y)),
> transform_body(G, nothing, M, R, []),
> transform_redo(M, R, S).
> S = [p(X), q(Y)]
>
> ?- G = (p(X), !, q(Y)),
> transform_body(G, nothing, M, R, []),
> transform_redo(M, R, S).
> S = ['$REDO'(_A), p(X), '$CUT'(_A), q(Y)]
>
> Introduced the new internal built-in '$REDO'/1 yesterday,
> together with '$SEQ/1 and '$ALT'/1. I guess it makes the
> cut inlinable, just inline the transformed stuff.
>
> This would mean inlining doesn't happen on the level
> of term/goal expansion, but a little lower. But before
> the encode, or maybe after the encoding?
>
> P.S.: I borrowed '$SEQ/1 and '$ALT'/1 from Jekejeke Prolog
> somehow, but the '$REDO'/1 and '$CUT'/1 stuff is
> a new old idea, specifically pursued in the Dogelog runtime,
>
> might later transfer back to Jekejeke Prolog, if successful
> in all respects. The last proof of concept will be providing
> if-then-else support based on '$REDO'/1 and '$CUT'/1.
> Mostowski Collapse schrieb am Freitag, 18. Juni 2021 um 13:08:06 UTC+2:
> > Why didn't O-Prolog have correct if-then-else? Don't
> > blame ISO. Currently we are working our way through
> > disjunction (;)/2 support for Dogelog runtime:
> >
> > Preview: Dogelog runtime can do Prolog disjunction. (Jekejeke)
> > https://gist.github.com/jburse/76c67a1529f306c9d376b3264a6a52b3#gistcomment-3784948
> >
> > Whats also planned support for if-then-else (->)/2. And
> > a new compilation scheme for '$ALT'/1, which we
> > didn't need in Jekejeke Prolog.

The new Android Store/Emulator inside Windows 11
is probably an anachronism, flogging a dead horse.

It seems JavaScript is everywhere, even Artificial
Intelligence becomes cookie cutting:

I built a Distance Matrix Routing App in React!
https://www.youtube.com/watch?v=43jfFU4FJZo

Mostowski Collapse schrieb am Dienstag, 22. Juni 2021 um 15:27:37 UTC+2:
> Since inlining works, Dogelog is now old enough for its
> own group. Dogelog got its own Facebook Group!
>
> https://www.facebook.com/groups/dogelog
> Mostowski Collapse schrieb am Freitag, 18. Juni 2021 um 13:58:21 UTC+2:
> > Now I have the feeling I can inline anything, even stuff
> > that has cut or whatever. Cut (!)/0 inside call/1 is now
> > translated as follows:
> >
> > ?- G = (p(X), q(Y)),
> > transform_body(G, nothing, M, R, []),
> > transform_redo(M, R, S).
> > S = [p(X), q(Y)]
> >
> > ?- G = (p(X), !, q(Y)),
> > transform_body(G, nothing, M, R, []),
> > transform_redo(M, R, S).
> > S = ['$REDO'(_A), p(X), '$CUT'(_A), q(Y)]
> >
> > Introduced the new internal built-in '$REDO'/1 yesterday,
> > together with '$SEQ/1 and '$ALT'/1. I guess it makes the
> > cut inlinable, just inline the transformed stuff.
> >
> > This would mean inlining doesn't happen on the level
> > of term/goal expansion, but a little lower. But before
> > the encode, or maybe after the encoding?
> >
> > P.S.: I borrowed '$SEQ/1 and '$ALT'/1 from Jekejeke Prolog
> > somehow, but the '$REDO'/1 and '$CUT'/1 stuff is
> > a new old idea, specifically pursued in the Dogelog runtime,
> >
> > might later transfer back to Jekejeke Prolog, if successful
> > in all respects. The last proof of concept will be providing
> > if-then-else support based on '$REDO'/1 and '$CUT'/1.
> > Mostowski Collapse schrieb am Freitag, 18. Juni 2021 um 13:08:06 UTC+2:
> > > Why didn't O-Prolog have correct if-then-else? Don't
> > > blame ISO. Currently we are working our way through
> > > disjunction (;)/2 support for Dogelog runtime:
> > >
> > > Preview: Dogelog runtime can do Prolog disjunction. (Jekejeke)
> > > https://gist.github.com/jburse/76c67a1529f306c9d376b3264a6a52b3#gistcomment-3784948
> > >
> > > Whats also planned support for if-then-else (->)/2. And
> > > a new compilation scheme for '$ALT'/1, which we
> > > didn't need in Jekejeke Prolog.

Something like this might do the job in the future,
and replace Android and iOS apps. Dunno?

https://www.w3.org/TR/miniapp-lifecycle/

Mostowski Collapse schrieb am Freitag, 25. Juni 2021 um 13:00:22 UTC+2:
> The new Android Store/Emulator inside Windows 11
> is probably an anachronism, flogging a dead horse.
>
> It seems JavaScript is everywhere, even Artificial
> Intelligence becomes cookie cutting:
>
> I built a Distance Matrix Routing App in React!
> https://www.youtube.com/watch?v=43jfFU4FJZo
> Mostowski Collapse schrieb am Dienstag, 22. Juni 2021 um 15:27:37 UTC+2:
> > Since inlining works, Dogelog is now old enough for its
> > own group. Dogelog got its own Facebook Group!
> >
> > https://www.facebook.com/groups/dogelog
> > Mostowski Collapse schrieb am Freitag, 18. Juni 2021 um 13:58:21 UTC+2:
> > > Now I have the feeling I can inline anything, even stuff
> > > that has cut or whatever. Cut (!)/0 inside call/1 is now
> > > translated as follows:
> > >
> > > ?- G = (p(X), q(Y)),
> > > transform_body(G, nothing, M, R, []),
> > > transform_redo(M, R, S).
> > > S = [p(X), q(Y)]
> > >
> > > ?- G = (p(X), !, q(Y)),
> > > transform_body(G, nothing, M, R, []),
> > > transform_redo(M, R, S).
> > > S = ['$REDO'(_A), p(X), '$CUT'(_A), q(Y)]
> > >
> > > Introduced the new internal built-in '$REDO'/1 yesterday,
> > > together with '$SEQ/1 and '$ALT'/1. I guess it makes the
> > > cut inlinable, just inline the transformed stuff.
> > >
> > > This would mean inlining doesn't happen on the level
> > > of term/goal expansion, but a little lower. But before
> > > the encode, or maybe after the encoding?
> > >
> > > P.S.: I borrowed '$SEQ/1 and '$ALT'/1 from Jekejeke Prolog
> > > somehow, but the '$REDO'/1 and '$CUT'/1 stuff is
> > > a new old idea, specifically pursued in the Dogelog runtime,
> > >
> > > might later transfer back to Jekejeke Prolog, if successful
> > > in all respects. The last proof of concept will be providing
> > > if-then-else support based on '$REDO'/1 and '$CUT'/1.
> > > Mostowski Collapse schrieb am Freitag, 18. Juni 2021 um 13:08:06 UTC+2:
> > > > Why didn't O-Prolog have correct if-then-else? Don't
> > > > blame ISO. Currently we are working our way through
> > > > disjunction (;)/2 support for Dogelog runtime:
> > > >
> > > > Preview: Dogelog runtime can do Prolog disjunction. (Jekejeke)
> > > > https://gist.github.com/jburse/76c67a1529f306c9d376b3264a6a52b3#gistcomment-3784948
> > > >
> > > > Whats also planned support for if-then-else (->)/2. And
> > > > a new compilation scheme for '$ALT'/1, which we
> > > > didn't need in Jekejeke Prolog.

Some more progress. Dogelog got a Prolog dynamic
database. We can now run this example and more:

:- dynamic(memo/2).
fib(N, X) :- memo(N, X), !.
fib(0, 0) :- !.
fib(1, 1) :- !.
fib(N, X) :- M is N-1, fib(M, Y), K is M-1,
fib(K, Z), X is Y+Z, assertz(memo(N, X)).
:- fib(10,X), write(X), nl.
:- fib(100,X), write(X), nl.

Dogelog will show me:

55
354224848179261915075

See also:

Preview: Dogelog got a Prolog dynamic database. (Jekejeke)
https://gist.github.com/jburse/ce2cc49168b637fb0472b3d958999c40#gistcomment-3798277

Pitty it doesn't work in TauProlog, not because of
lack of dynamic database, TauProlog has dynamic database,
but bigint is missing. This is what TauProlog shows me:

?- fib(100,X).
uncaught exception: error(evaluation_error(int_overflow),is/2)

On Thursday, July 1, 2021 at 1:36:12 AM UTC+10, burs...@gmail.com wrote:
> Some more progress. Dogelog got a Prolog dynamic
> database. We can now run this example and more:
>
> :- dynamic(memo/2).
> fib(N, X) :- memo(N, X), !.
> fib(0, 0) :- !.
> fib(1, 1) :- !.
> fib(N, X) :- M is N-1, fib(M, Y), K is M-1,
> fib(K, Z), X is Y+Z, assertz(memo(N, X)).
> :- fib(10,X), write(X), nl.
> :- fib(100,X), write(X), nl.
>
> Dogelog will show me:

my FIB function went into an infinite loop

R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R

fib1 2

YES

18 fib1 2
N=2
R fib1 2

18 1 if
18 2 bigger N 0
18 3 clear M
18 4 sub M N 1
18 5 fib1 M S
18 fib1 1 S
N=1
fib1 1 S

18 1 if
18 2 bigger N 0
18 3 clear M
18 4 sub M N 1
18 5 fib1 M S
18 fib1 M S
fib1 M S

18 1 if
18 2 bigger N 0
18 3 clear M
18 4 sub M N 1
18 5 fib1 M S
18 fib1 M S
fib1 M S

18 1 if
18 2 bigger N 0
18 3 clear M
18 4 sub M N 1
18 5 fib1 M S
18 fib1 M S
fib1 M S

18 1 if
18 2 bigger N 0
18 3 clear M
18 4 sub M N 1
18 5 fib1 M S
18 fib1 M S
fib1 M S

18 1 if
18 2 bigger N 0
18 3 clear M
18 4 sub M N 1
18 5 fib1 M S
18 fib1 M S
fib1 M S

18 1 if
18 2 bigger N 0
18 3 clear M
18 4 sub M N 1
18 5 fib1 M S
18 fib1 M S
fib1 M S

More flop than fib.

Graham Cooper schrieb am Mittwoch, 30. Juni 2021 um 23:09:01 UTC+2:
> On Thursday, July 1, 2021 at 1:36:12 AM UTC+10, burs...@gmail.com wrote:
> > Some more progress. Dogelog got a Prolog dynamic
> > database. We can now run this example and more:
> >
> > :- dynamic(memo/2).
> > fib(N, X) :- memo(N, X), !.
> > fib(0, 0) :- !.
> > fib(1, 1) :- !.
> > fib(N, X) :- M is N-1, fib(M, Y), K is M-1,
> > fib(K, Z), X is Y+Z, assertz(memo(N, X)).
> > :- fib(10,X), write(X), nl.
> > :- fib(100,X), write(X), nl.
> >
> > Dogelog will show me:
> my FIB function went into an infinite loop
>
>
>
>
>
>
> R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R
>
> fib1 2
>
> YES
>
>
>
> 18 fib1 2
> N=2
> R=
> fib1 2
>
> 18 1 if
> 18 2 bigger N 0
> 18 3 clear M
> 18 4 sub M N 1
> 18 5 fib1 M S
> 18 fib1 1 S
> N=1
> fib1 1 S
>
> 18 1 if
> 18 2 bigger N 0
> 18 3 clear M
> 18 4 sub M N 1
> 18 5 fib1 M S
> 18 fib1 M S
> fib1 M S
>
> 18 1 if
> 18 2 bigger N 0
> 18 3 clear M
> 18 4 sub M N 1
> 18 5 fib1 M S
> 18 fib1 M S
> fib1 M S
>
> 18 1 if
> 18 2 bigger N 0
> 18 3 clear M
> 18 4 sub M N 1
> 18 5 fib1 M S
> 18 fib1 M S
> fib1 M S
>
> 18 1 if
> 18 2 bigger N 0
> 18 3 clear M
> 18 4 sub M N 1
> 18 5 fib1 M S
> 18 fib1 M S
> fib1 M S
>
> 18 1 if
> 18 2 bigger N 0
> 18 3 clear M
> 18 4 sub M N 1
> 18 5 fib1 M S
> 18 fib1 M S
> fib1 M S
>
> 18 1 if
> 18 2 bigger N 0
> 18 3 clear M
> 18 4 sub M N 1
> 18 5 fib1 M S
> 18 fib1 M S
> fib1 M S

On Thursday, July 1, 2021 at 11:58:10 AM UTC+10, burs...@gmail.com wrote:
> More flop than fib.

fibb M N L
-if
-bigger L 0
-add S M N
-print S
-sub L L 1
-fibb N S L
-endif

1 2 3 5 8 13 21 34 55 89

fibb 0 1 10 [?]

YES