• ANN: Dogelog Player 2.3.2 (Schwartzian Transform)

    From Mild Shock@janburse@fastmail.fm to comp.lang.javascript on Mon Sep 28 13:04:30 2026
    From Newsgroup: comp.lang.javascript

    Dear All,

    We are happy to announce a new
    edition of the Dogelog Player:

    - Ehanced common:
    Triggered by SWI-Prologs new built-ins,
    we did a revision of library(cyclic) now
    exposing term_minimal/2 and term_decompose/2.
    Turns out we could also speed-up distinct/1
    and aggregate/3, by using term_minimal/2
    alone in the deduplication phase. And
    then for aggregate/3 deploying a Schwartzian
    Transform with term_decompose/2 in
    the sorting phase.

    - Enhanced top-level:
    We also did a revision of the top-level.
    For example answer substitutions are now
    shown more compact, so that a query X = f(A,Y),
    Y = f(B,X) can keep its shape. Further
    the debugger now uses the visibility rules
    of listing/[0,1]. Even overriding the
    provenance of goals arguments and only
    prompting in case they branch back
    into user clauses.

    - New undo/1:
    Realizing the analogue to our recent rogue
    '$CUT'/1 for cleanup goals, this time not
    scanning the choice points but the variable
    trail to find undo goals, we could arrive
    at a undo/1. Because of atomicity concerns,
    its not directly exposed, and subsequently
    the user facing API has now new set_arg/3
    and setup_undo/2 only.

    Have Fun!

    Jan Burse, September 28, 2026, https://www.herbrand.ai/
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Mild Shock@janburse@fastmail.fm to comp.lang.javascript on Tue Oct 6 10:08:54 2026
    From Newsgroup: comp.lang.javascript

    Hi,

    In the following we show that P(purity) has
    currently almost zero probability, i.e that a
    Prolog system will provide the first pillar of
    purity, i.e. the dif/2 constraint. Even if a
    Prolog system can deal with a large number of
    dif/2 constraints, it might still fail
    the reclamation test.

    Testing Ulrich Neumerkels maplist(dif) problem
    with a time limit of 1000 ms, we found Trealla
    Prolog, EyeProlog and Scryer Prolog only
    approached N=1000. While SWI-Prolog and Dogelog
    Player climbed above N=1000000. Further a garbage
    collection test was only survived by
    Dogelog Player.

    Bye

    See also:

    The implementation of dif/2 with Open Trees https://medium.com/2989/6e2914596a0f

    Mild Shock schrieb:
    Dear All,

    We are happy to announce a new
    edition of the Dogelog Player:

    - Ehanced common:
    Triggered by SWI-Prologs new built-ins,
    we did a revision of library(cyclic) now
    exposing term_minimal/2 and term_decompose/2.
    Turns out we could also speed-up distinct/1
    and aggregate/3, by using term_minimal/2
    alone in the deduplication phase. And
    then for aggregate/3 deploying a Schwartzian
    Transform with term_decompose/2 in
    the sorting phase.

    - Enhanced top-level:
    We also did a revision of the top-level.
    For example answer substitutions are now
    shown more compact, so that a query X = f(A,Y),
    Y = f(B,X) can keep its shape. Further
    the debugger now uses the visibility rules
    of listing/[0,1]. Even overriding the
    provenance of goals arguments and only
    prompting in case they branch back
    into user clauses.

    - New undo/1:
    Realizing the analogue to our recent rogue
    '$CUT'/1 for cleanup goals, this time not
    scanning the choice points but the variable
    trail to find undo goals, we could arrive
    at a undo/1. Because of atomicity concerns,
    its not directly exposed, and subsequently
    the user facing API has now new set_arg/3
    and setup_undo/2 only.

    Have Fun!

    Jan Burse, September 28, 2026, https://www.herbrand.ai/

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Mild Shock@janburse@fastmail.fm to comp.lang.javascript on Fri Oct 16 00:38:39 2026
    From Newsgroup: comp.lang.javascript

    Hi,

    Dogelog Player is a Prolog system for the
    JavaScript, Python and Java targets. It is mostly
    written in 100% Prolog. Our garbage collector
    already used GC frequency autotuning via user_time.
    We added a GC generation choice autotuning
    feature using user_memory.

    The heuristic is to use 75% of max_memory as
    a limit for user_memory to trigger an emergency
    major mark & sweep. The Dogelog Player can now
    run loop/1 indefinitely. Less welcome is the
    performance drop from 30 to 10 MLips. SWI-Prolog
    does not show the same degradation.

    Bye

    See also:

    Dogelog Player: Heap Pressure for Garbage Collection https://medium.com/2989/5791ef8834d5

    Mild Shock schrieb:
    Hi,

    In the following we show that P(purity) has
    currently almost zero probability, i.e that a
    Prolog system will provide the first pillar of
    purity, i.e. the dif/2 constraint. Even if a
    Prolog system can deal with a large number of
    dif/2 constraints, it might still fail
    the reclamation test.

    Testing Ulrich Neumerkels maplist(dif) problem
    with a time limit of 1000 ms, we found Trealla
    Prolog, EyeProlog and Scryer Prolog only
    approached N=1000. While SWI-Prolog and Dogelog
    Player climbed above N=1000000. Further a garbage
    collection test was only survived by
    Dogelog Player.

    Bye

    See also:

    The implementation of dif/2 with Open Trees https://medium.com/2989/6e2914596a0f

    Mild Shock schrieb:
    Dear All,

    We are happy to announce a new
    edition of the Dogelog Player:

    - Ehanced common:
    Triggered by SWI-Prologs new built-ins,
    we did a revision of library(cyclic) now
    exposing term_minimal/2 and term_decompose/2.
    Turns out we could also speed-up distinct/1
    and aggregate/3, by using term_minimal/2
    alone in the deduplication phase. And
    then for aggregate/3 deploying a Schwartzian
    Transform with term_decompose/2 in
    the sorting phase.

    - Enhanced top-level:
    We also did a revision of the top-level.
    For example answer substitutions are now
    shown more compact, so that a query X = f(A,Y),
    Y = f(B,X) can keep its shape. Further
    the debugger now uses the visibility rules
    of listing/[0,1]. Even overriding the
    provenance of goals arguments and only
    prompting in case they branch back
    into user clauses.

    - New undo/1:
    Realizing the analogue to our recent rogue
    '$CUT'/1 for cleanup goals, this time not
    scanning the choice points but the variable
    trail to find undo goals, we could arrive
    at a undo/1. Because of atomicity concerns,
    its not directly exposed, and subsequently
    the user facing API has now new set_arg/3
    and setup_undo/2 only.

    Have Fun!

    Jan Burse, September 28, 2026, https://www.herbrand.ai/


    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Mild Shock@janburse@fastmail.fm to comp.lang.javascript on Fri Oct 23 01:41:48 2026
    From Newsgroup: comp.lang.javascript

    Hi,

    Prolog systems with attribute variables saw growing
    use cases for constraint logic programming with
    implementations such as CLP(FD), CLP(B), etc.. for
    example in SWI-Prolog. But one might wonder why
    they are also associated with concurrent
    logic programming.

    The new when/2 primitive allows scheduling goals
    on multiple events. By modelling eating forks via
    lazy lists, we could replicated the Dining Philosopher
    problem. The lazy lists need to be scheduled by an
    additional program and we can show that such a
    scheduler can run forever.

    Bye

    See also:

    Dogelog Player: Dining Philosophers with when/2 https://medium.com/2989/bc3103c2d544

    Mild Shock schrieb:
    Hi,

    Dogelog Player is a Prolog system for the
    JavaScript, Python and Java targets. It is mostly
    written in 100% Prolog. Our garbage collector
    already used GC frequency autotuning via user_time.
    We added a GC generation choice autotuning
    feature using user_memory.

    The heuristic is to use 75% of max_memory as
    a limit for user_memory to trigger an emergency
    major mark & sweep. The Dogelog Player can now
    run loop/1 indefinitely. Less welcome is the
    performance drop from 30 to 10 MLips. SWI-Prolog
    does not show the same degradation.

    Bye

    See also:

    Dogelog Player: Heap Pressure for Garbage Collection https://medium.com/2989/5791ef8834d5

    Mild Shock schrieb:
    Hi,

    In the following we show that P(purity) has
    currently almost zero probability, i.e that a
    Prolog system will provide the first pillar of
    purity, i.e. the dif/2 constraint. Even if a
    Prolog system can deal with a large number of
    dif/2 constraints, it might still fail
    the reclamation test.

    Testing Ulrich Neumerkels maplist(dif) problem
    with a time limit of 1000 ms, we found Trealla
    Prolog, EyeProlog and Scryer Prolog only
    approached N=1000. While SWI-Prolog and Dogelog
    Player climbed above N=1000000. Further a garbage
    collection test was only survived by
    Dogelog Player.

    Bye

    See also:

    The implementation of dif/2 with Open Trees
    https://medium.com/2989/6e2914596a0f

    Mild Shock schrieb:
    Dear All,

    We are happy to announce a new
    edition of the Dogelog Player:

    - Ehanced common:
    Triggered by SWI-Prologs new built-ins,
    we did a revision of library(cyclic) now
    exposing term_minimal/2 and term_decompose/2.
    Turns out we could also speed-up distinct/1
    and aggregate/3, by using term_minimal/2
    alone in the deduplication phase. And
    then for aggregate/3 deploying a Schwartzian
    Transform with term_decompose/2 in
    the sorting phase.

    - Enhanced top-level:
    We also did a revision of the top-level.
    For example answer substitutions are now
    shown more compact, so that a query X = f(A,Y),
    Y = f(B,X) can keep its shape. Further
    the debugger now uses the visibility rules
    of listing/[0,1]. Even overriding the
    provenance of goals arguments and only
    prompting in case they branch back
    into user clauses.

    - New undo/1:
    Realizing the analogue to our recent rogue
    '$CUT'/1 for cleanup goals, this time not
    scanning the choice points but the variable
    trail to find undo goals, we could arrive
    at a undo/1. Because of atomicity concerns,
    its not directly exposed, and subsequently
    the user facing API has now new set_arg/3
    and setup_undo/2 only.

    Have Fun!

    Jan Burse, September 28, 2026, https://www.herbrand.ai/



    --- Synchronet 3.22a-Linux NewsLink 1.2