Re: MPS: User GC customizations

[Gerd Möllmann]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Gerd Möllmann <gerd.moellmann@gmail.com>
>> Cc: Pip Cet <pipcet@protonmail.com>,  Eli Zaretskii <eliz@gnu.org>,
>>   emacs-devel@gnu.org,  eller.helmut@gmail.com
>> Date: Thu, 04 Jul 2024 14:51:16 +0200
>> 
>> >>   igc-step-interval is a variable defined in ‘C source code’.
>> >>
>> >>   Its value is 0.1
>> >>
>> >>   How much time is MPS allowed to spend in GC when Emacs is idle.
>> >>   The value is in seconds, and should be a non-negative number.  It can
>> >>   be either an integer or a float.  The default value is 0 which means .
>> >>   don’t do something when idle.  Negative values and values that are not 
>> >> numbers
>> >>   are handled as if they were the default value.
>> >>
>> >> It's for interactive use.
>> > [...]
>> What this variable does is give MPS notice that the client is currently
>> idle and it might be a good time to do some work.
>
> Is that really what this variable does?  My reading of the
> documentation is that it tells MPS how soon to stop GC which it
> started when Emacs was idle.  IOW, it's not about _triggering_ GC,
> it's about _ending_ it.  And the reason, AFAIU, is to avoid making
> Emacs responses slow because MPS started GC when it detected that
> Emacs is idle.

I'm taking my view from this;:

  2.4.6 Using idle time for collection
  ------------------------------------

  Some types of program have “idle time” in which they are waiting for an
  external event such as user input or network activity.  The MPS provides
  a function, *note mps_arena_step(): 19c, for making use of idle time to
  make memory management progress.
  ...
  
The name interval I've taken from 

 -- C Function: *note mps_bool_t: 129. mps_arena_step (mps_arena_t
          arena, double interval, double multiplier)

     ‘interval’ is the time, in seconds, the MPS is permitted to take.
     It must not be negative, but may be ‘0.0’.

Alas, it doesn't promise to always obey interval:

     The MPS makes every effort to
     return from this function within ‘interval’ seconds, but cannot
     guarantee to do so, as it may need to call your own scanning code.
     It uses ‘multiplier’ to decide whether to commence long-duration
     operations that consume CPU (such as a full collection): it will
     only start such an operation if it is expected to be completed
     within ‘multiplier * interval’ seconds.

(Haven't found out yet to prevent full collections with a message "client
predicts plenty of idle time" or some such yet. Please see the comment
in igc.c where mps_arena_step is called.)