HackMD
The snake of Theseus - Pablo Galindo Salgado
Collaborative writing start from below
從這裡開始共筆
Slido
This talk is about what makes python feel like python.
01 - the ship of theseus
philosophy dilemma: If you replace all part of this ship, it is still the same ship?
as we changed python as we remove or add up things
(code survive over time)
we don't changed things (it's the same ship)
python is really written by a lot of people,… the things become more complicated.
Python is changed quite quickly.
Python is actually does not exist.
(radio waves metaphor. You can not see the pink color on the spectrum. It's combanation of different photons)
different people have different ideas
Python only exists in our mind
lives in minds
02 - The Grammar
who says what is python and what is not - it's grammar
(rule_names example)
metagrammar: grammars describe grammars
describe what is python
(expanded rules)
LL1-grammar
simple and restrict
idomtic simple to write, coded in LL1
reference:Full Grammar specification
-
Left to right
parses input from left to right -
Leftmost derivation
-
1 Token Lookahead
Biggest restriciton
Grammar restriction
(break example)
is something continue could be a simple statement?
(ambiguous rules)
ars est quelare artem
it's importment that when you creating things do not get in the middle
(if you write after a grammar, it's ambiguous/complicated)
- nonterminal in the grammar…
- to make it run in liner
memoization.. - context in grammars
1st python 1990
PEG parsers have infinite backtracking
https://peps.python.org/pep-0617/
03 A New Garmmar World
in python sometimes we are introducing new keywords(ideas)
hard keywords: can not be used as varibles
soft keywords: contextual, some words can be used as variables(e.g. await, async in py3.4, 3.5), backward compatible
(e.g the use of underscore)
pep636 structure
only possible with the new peg parser
Python 3.10 release note:
- This new syntax uses the non LL(1) capacities of the new parser. Check PEP 617 for more details. PEP 617 – New PEG parser for CPython
04 The Storm of Tpying
the two main topics
- static typing
- is used by tools analysing the source code without excuting it (e.g. mypy)
- Dynamic typing
- is examine at runtime to produce some effects (e.g. Pydantic)
Did you know this?
- PEP 484
- Introduced the typing lang with syntax, annotations, the typing module and other infrastructure.
- PEP 563
- Introduced
from __future__ import annotations
- Introduced
- PEP 649
- Expands the effect of
from __future__ import annotations
- Expands the effect of
how users see annotations
-
small projects and education
-
enterprise and large projects
PEP 649 is accepted for Python 3.12
05 battling the GIL
- OCT.7 2021
- NOGIL Python targeting 3.11
- JAN 9 2023
- PEP 703 - tentatively accepted for Python 3.13
- July 28 2024
Challenges of nogic
-
maintanance: increase the Python team's maintenance bourden
-
compatibility: isses for C extensions
-
performance
NOGIL -
packaging: for a while packages might need to be released in two versions (as GIL and NOGIL), which complicates releases and testing too
20
Types of programs
I/O bounded code: doesn't quite benefit from NOGIL
CPU bounded code: does benefit from NOGIL if there's parallelism
Mixed workload
Timeline of adoption
01 0-3 years : experimental builds
02 3-5 years : suppported builds
03 5+ years : NOGIL is the only build of Python
Challengs for c extensions
global state
debugging is harder
implicit gil as lock
interoperabitpality
looking infrastructure
validation
The Snake of Theseus
@pyblogsal
- what is the biggest difference
it has to be something
-
as a thought experiment
-
can you share us one mistake you've made as a CPython
change from [] to main
. would you consider
. how would you suggest for
- how to consolidate everyone's understanding of python
fantastic question, it's hard
. what's the hardest
Below is the part that speaker updated the talk/tutorial after speech
講者於演講後有更新或勘誤投影片的部份
