Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.javascript > #124488

Dogelog Player: Heap Pressure for Garbage Collection (Was: The implementation of dif/2 with Open Trees)

Show key headers only | View raw

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/
> 

Back to comp.lang.javascript | Previous — Previous in thread | Find similar | Unroll thread

Thread

ANN: Dogelog Player 2.3.2 (Schwartzian Transform) Mild Shock <janburse@fastmail.fm> - 2026-09-28 13:04 +0200
  The implementation of dif/2 with Open Trees (Was: ANN: Dogelog Player 2.3.2 (Schwartzian Transform)) Mild Shock <janburse@fastmail.fm> - 2026-10-06 10:08 +0200
    Dogelog Player: Heap Pressure for Garbage Collection (Was: The implementation of dif/2 with Open Trees) Mild Shock <janburse@fastmail.fm> - 2026-10-16 00:38 +0200

csiph-web