posita / dyce Goto Github PK

To do

• Study feasibility of adapting mike to GitLab Pages
• Set up CI on GitLab
• Turn off mirroring from this repo to the GitLab one
• Update helpers/draft-release.sh and docs via commit in GitLab repo to point to gitlab.com/gitlab.io instead of github.com/github.io
  • grep 2>|/dev/null -il github * */* for places to look
• Set up reverse mirroring or replace this repo with a link to the GitLab one

From here (scroll down):

Related to this shit.

When trying to execute a program that doesn't use a terminal (.pyw extension) os.get_terminal_size() throws "ValueError: bad file descriptor".
Resolved substituting it with shutil.get_terminal_size(), if you want to fix it.

Currently, Rollers support relatively low-level operations, the most complicated being selection. That leaves implementing things like "exploding" dice up to library clients. Histograms already support exploding dice through substitution. Rollers should implement an analogous mechanism.

I'm currently working on a project¹ for analyzing games at a higher level (e.g. expected outcomes for Warhammer attack sequences), for which I have been using Pierre Denis's lea package ² for the low-level probability analysis. I ran across your work³ and am now evaluating whether it's better suited to my use case. In any event, I thought you might also be interested in evaluating lea as it seems to be an excellent low-level library for discrete probability analysis, although it lacks some amenities like type hints.

This isn't so much an "issue" as "hello, I'm working in a related space, here's what I've found so far." Cheers!

Here's a contrived example for which dyce doesn't provide a convenient interface:

1. Start with a total of zero.
2. Roll a six-sided die. If the face was a six, go to step 3 (i.e., replace with whatever that produces). Otherwise add the face to the total and stop.
3. Roll a four-sided die. Add the face to the total. If the face was a one, go to step 2. Otherwise stop.

Maybe the ability for expand to return an optional coalesce override would work?

from operator import __add__
from dyce import H
from dyce.h import coalesce_replace

def expand(h: H, outcome):
  if h == d6:
    if outcome == 6:
      return d4  # uses default passed to substitute
    else:
      return outcome, __add__  # overrides coalesce for this phase
  elif h == d4:
    if outcome == 1:
      return d6, lambda: h, __: h + outcome  # like coalesce_replace, but adds the outcome from the d4
    else:
      return outcome, __add__
  else:
    assert False, f"unrecognized histogram {h}"

H(6).substitute(expand, coalesce_replace, max_depth=4)