niklasf
Code refactor, meta stuff
- [ ] Fix inconsistent string formatting (both
.format()and f-string methods are used) - [ ] Refactor & optimize algorithms (syzygy, gaviota)
- [ ] Shorten maximum line length to 99 (thank later)
- [ ] Migrate to PEP 604 X | Y type syntax due to #564
- [ ] Add a requirements.txt file (or even better - migrate from setuptools to poetry)
- [ ] Fix 517 issues reported by Ruff, possibly add a Ruff spec file or add a proper section in pyproject.toml if we're into poetry
- [ ] Bump pylint rating, my initial goal: 9.00/10, current rating: 7.64/10
Some algorithms do need logic refactoring, for instance
https://github.com/niklasf/python-chess/blob/f76c387ef9c68555a08b1629b56655d85f86d949/chess/syzygy.py#L323-L352
which could simply be truncated into
for i in range(5):
j = 0
a = PFACTOR[i]
p = PAWNIDX[i]
for k in range(4):
s = 0
t = 6 * (k+1)
while j < t:
p[j] = s
s += 1 if i == 0 else binom(PTWIST[INVFLAP[j]], i)
j += 1
a[k] = s
I'll work on those enhancements soon. Please let me know what you think about migrating to poetry and maybe possibly migrating tests to pytest to make them easier to read and work with. @niklasf
Hi, some thoughts about these:
- (1), (2), (3): Sure, if the new code would be more readable. For (2), the code was ported line by line from a C original. It has since diverged, so I'd be fine with giving up the 1:1 correspondence and make it look more like Python code.
- (4) will only be merged after Python 3.9 is past its end of life.
- (5): There are no dependencies :)
- (6), (7): Sure, if the findings make sense.
For pytest, at a glance I don't see how it would improve dx for this project. Do you have any particular features in mind?
About (4), I think as long as you import from __future__ importannotations, you should be able to use the new syntax just fine, we use it on berserk and support 3.8
It can actually work even for type hints evaluated at runtime since https://github.com/alexmojaki/eval_type_backport.