A “frozen” dictionary for Python

A “frozen” dictionary for Python(lwn.net)

191 points by jwilk 158 days ago | 167 comments

pansa2 157 days ago |

I wonder whether Raymond Hettinger has an opinion on this PEP. A long time ago, he wrote: "freezing dicts is a can of worms and not especially useful".

https://mail.python.org/pipermail/python-dev/2006-February/0...

pkulak 157 days ago | |

This is why I love how Rust approached this; almost by accident to make borrow checking work. Every reference is either mutable or not, and (with safe code), you can't use an immutable reference to get a mutable reference anywhere down the chain. So you can slowly construct a map through a mutable reference, but then return it out of a function as immutable, and that's the end of it. It's no longer ever mutable, and no key or value is either. There's no need to make a whole new object called FrozenHashMap, and then FrozenList, and FrozenSet, etc. You don't need a StringBuilder because String is mutable, unless you don't want it to be. It's all just part of the language.

Kotlin _kinda_ does this as well, but if you have a reference to an immutable map in Kotlin, you are still free to mutate the values (and even keys!) as much as you like.

rcxdude 157 days ago | | |

Only if you're returning a reference or wrapping it in something that will only ever return a reference. If you return an object by value ('owned'), then you can do what you like with it and 'mut' is just an light guardrail on that particular name for it.

vlovich123 157 days ago | | |

You cannot return an immutable version. You can return it owned (in which case you can assign/reassign it to a mut variable at any point) or you can take a mut reference and return an immutable reference - but whoever is the owner can almost always access it mutably.

the__alchemist 157 days ago | | |

My favorite part of rust is its explicit control over mutability, in the manner you describe.

jonathaneunice 157 days ago | |

That's a great link and recommended reading.

It explains a lot about the design of Python container classes, and the boundaries of polymorphism / duck typing with them, and mutation between them.

I don't always agree with the choices made in Python's container APIs...but I always want to understand them as well as possible.

Also worth noting that understanding changes over time. Remember when GvR and the rest of the core developers argued adamantly against ordered dictionaries? Haha! Good times! Thank goodness their first wave of understanding wasn't their last. Concurrency and parallelism in Python was a TINY issue in 2006, but at the forefront of Python evolution these days. And immutability has come a long way as a design theme, even for languages that fully embrace stateful change.

zahlman 157 days ago | | |

> Also worth noting that understanding changes over time. Remember when GvR and the rest of the core developers argued adamantly against ordered dictionaries? Haha! Good times!

The new implementation has saved space, but there are opportunities to save more space (specifically after deleting keys) that they've now denied themselves by offering the ordering guarantee.

mvanbaak 157 days ago | |

This was 19 (almost) 20 years ago. As stated in the lwn.net article, a lot of concurrency has been added to python, and it might now be time for something like a frozendict.

Things that were not useful in 2006 might be totally useful in 2026 ;P

Still, like you, I'm curious wether he has anything to say about it.

aewens 157 days ago | | |

I think Raymond Hettinger is called out specially here because he did a well known talk called [Modern Dictionaries](https://youtu.be/p33CVV29OG8) where around 32:00 to 35:00 in he makes the quip about how younger developers think they need new data structures to handle new problems, but eventually just end up recreating / rediscovering solutions from the 1960s.

“What has been is what will be, and what has been done is what will be done; there is nothing new under the sun.”

dkarl 157 days ago | |

It's interesting that he concludes that freezing dicts is "not especially useful" after addressing only a single motivation: the use of a dictionary as a key.

He doesn't address the reason that most of us in 2025 immediately think of, which is that it's easier to reason about code if you know that certain values can't change after they're created.

What a change in culture over the last 20 years!

morshu9001 157 days ago | | |

You can't really tell though. Maybe the dict is frozen but the values inside aren't. C++ tried to handle this with constness, but that has its own caveats that make some people argue against using it.

eru 154 days ago | |

Ha, Raymond Hettinger has a lot of opinions. Great guy and I admire his dedication to Python, but in my own experience (and the experience of some other contributors), he has a chilling effect on contributions to certain parts of the CPython code base.

Not that it's entirely unwarranted, of course.

zahlman 157 days ago | |

> Another PEP 351 world view is that tuples can serve as frozenlists; however, that view represents a Liskov violation (tuples don't support the same methods). This idea resurfaces and has be shot down again every few months.

... Well, yes; it doesn't support the methods for mutation. Thinking of ImmutableFoo as a subclass of Foo is never going to work. And, indeed, `set` and `frozenset` don't have an inheritance relationship.

I normally find Hettinger very insightful so this one is disappointing. But nobody's perfect, and we change over time (and so do the underlying conditions). I've felt like frozendict was missing for a long time, though. And really I think the language would have been better with a more formal concept of immutability (e.g. linking it more explicitly to hashability; having explicit recognition of "cache" attributes, ...), even if it didn't go the immutable-by-default route.

kccqzy 157 days ago | | |

Apple (or perhaps NeXT) has solved this problem already in Objective-C. Look at NSArray and NSMutableArray, or NSData and NSMutableData. It’s intuitive and Liskov-correct to make the mutable version a subclass of the immutable version. And it’s clearly wrong to have the subclass relationship the other way around.

Given how dynamic Python is, such a subclass relationship need not be evident at the C level. You can totally make one class whose implementation is independent of another class a subclass of the other, using PEP 3119. This gives implementations complete flexibility in how to implement the class while retaining the ontological subclass relationship.

pansa2 157 days ago | | |

> ImmutableFoo as a subclass of Foo is never going to work. And, indeed, `set` and `frozenset` don't have an inheritance relationship.

Theoretically, could `set` be a subclass of `frozenset` (and `dict` of `frozendict`)? Do other languages take that approach?

> linking [immutability] more explicitly to hashability

AFAIK immutability and hashability are equivalent for the language's "core" types. Would it be possible to enforce that equivalence for user-defined types, given that mutability and the implementation of `__hash__` are entirely controlled by the programmer?

immibis 157 days ago | | |

ImmutableFoo can't be a subclass of Foo, since it loses the mutator methods. But nor can Foo be a subclass of ImmutableFoo, since it loses the axiom of immutability (e.g. thread-safety) that ImmutableFoo has.

When you interpret Liskov substitution properly, it's very rare that anything Liskov-substitutes anything, making the entire property meaningless. So just do things based on what works best in the real world and aim for as much Liskov-substitution as is reasonable. Python is duck-typed anyway.

It's a decent guiding principle - Set and ImmutableSet are more substitutable than Set and Map, so Set deriving from ImmutableSet makes more sense than Set deriving from Map. It's just not something you can ever actually achieve.

FreakLegion 157 days ago | | |

> I've felt like frozendict was missing for a long time, though.

Type the dict as a mapping when you want immutability:

  x: Mapping[int, int] = {1: 1}

  x[1] = 2  # Unsupported target for indexed assignment ("Mapping[int, int]").

The only problem I've seen with this is:

  y = {}
  y[x] = 0  # Mypy thinks this is fine. Mapping is hashable, after all!

The issue here is less that dict isn't hashable than that Mapping is, though.

tmp10423288442 157 days ago | | |

And, to the point of this proposal, `dict` and `frozendict` don't have an inheritance relationship either.

ndr 157 days ago | |

Immutability it's a joy to work with. Ask anyone who's worked with Clojure's dicts.

morshu9001 157 days ago | |

I agree, same with frozenset. If you really want to use one of those as a key, convert to a tuple. There might be niche use cases for all this, but it's not something that the language or even the standard lib need to support.

boothby 157 days ago | | |

Problem being that sets aren't consistently ordered and conversion to a tuple can result in an exponential (specifically, factorial) explosion in the number of possible keys associated with a single set. Nor can you sort all objects. Safe conversion of sets to tuples for use as keys is possible but the only technique I know requires an auxiliary store of objects (mapping objects to the order in which they were first observed), which doesn't parallelize well.

perrygeo 157 days ago |

Python discovers immutable data, and gets it wrong. Frozendict is a blunt instrument - instead of a mutable free-for-all, it just locks the data for the entire lifecycle. Brace for the wave of code littered with deep copies, proudly proclaiming how functional it all is.

If you want real immutable data structures, not a cheap imitation, check out pyrsistent.

polyrand 157 days ago |

A frozen dictionary would be very welcome. You can already do something similar using MappingProxyType [0]

  from types import MappingProxyType
  
  d = {}
  
  d["a"] = 1
  d["b"] = 2
  
  print(d)
  
  frozen = MappingProxyType(d)
  
  print(frozen["a"])
  
  # Error:
  frozen["b"] = "new"

[0]: https://docs.python.org/3/library/types.html#types.MappingPr...

zahlman 157 days ago | |

> You can already do something similar

Only if you deny access to the underlying real dict.

ali_m 157 days ago | | |

Yes, this only prevents the callee from mutating it, it can't provide a strong guarantee that the underlying mapping won't be changed upstream (and hence MappingProxyType can't be washable).

code_biologist 157 days ago |

Shoutout to pyrsistent, a personal favorite library for frozen/immutable/hashable data structures whenever I need to use lists, sets, or dicts as dict keys, or make sets of such items: https://github.com/tobgu/pyrsistent

sevensor 157 days ago |

Concurrency is a good motivation, but this is super useful even in straight line code. There’s a huge difference between functions that might mutate a dictionary you pass in to them and functions that definitely won’t. Using Mapping is great, but it’s a shallow guarantee because you can violate it at run time.

quietbritishjim 156 days ago | |

> There’s a huge difference between functions that might mutate a dictionary you pass in to them and functions that definitely won’t.

Maybe I misunderstood, but it sounds to me like you're hoping for the following code to work:

   def will_not_modify_arg(x: frozendict) -> Result:
       ...
   
   foo = {"a": 1, "b": 2}  # type of foo is dict
   r = will_not_modify_arg(foo)

But this won't work (as in, type checkers will complain) because dict is not derived from frozendict (or vice-versa). You'd have to create a copy of the dict to pass it to the function. (Aside from presumably not being what you intended, you can already do that with regular dictionaries to guarantee the original won't change.)

sevensor 156 days ago | | |

No, I’d type the function argument as a Mapping. Frozendict is so that the function will raise an exception if it violates its type signature.

Edit: that is, if as the caller you want foo to be immutable, then you make it a frozendict

sundarurfriend 157 days ago |

Can someone ELI5 the core difference between this and named tuples, for someone who is not deep into Python? ChatGPT's answer boiled down to: unordered (this) vs ordered (NTs), "arbitrary keys, decided at runtime" vs "fixed set of fields decided at definition time" (can't an NT's keys also be interpolated from runtime values?), and a different API (`.keys()`, `.items()`), etc (I'm just giving this as context btw, no idea if there's inaccuracies in these).

So could this also have been approached from the other side, as in making unordered NamedTuples with support for the Mapping API? The line between dictionaries and named tuples and structs (across various languages) has always seemed a bit blurry to me, so I'm trying to get a better picture of it all through this.

rurban 157 days ago |

> so having a safe way to share dictionaries between threads will be a boon

Since only the keys are const, the values not, frozendict is not thread-safe per se. There needs to be a small lock around the value getter and setter.

varelaz 157 days ago | |

it's thread safe on operations on the dict but not on the values. Same relates to other immutable structures like tuples. Lock will not help here cause unsafety comes from operation on value after value is obtained.

zzzeek 157 days ago |

SQLAlchemy has its own frozendict which we've had in use for many years, we have it as a pretty well performing cython implementation these days, and I use it ubiquitously. Would be a very welcome addition to the stdlib.

This proposal is important enough that I chimed in on the thread with a detailed example of how SQLAlchemy uses this pattern:

https://discuss.python.org/t/pep-814-add-frozendict-built-in...

bilsbie 157 days ago |

Wow weird Mandela effect for me. I really remember this being a built and actually using it.

aewens 157 days ago | |

You may be thinking of the `frozenset()` built in or the third party Python module [frozendict](https://pypi.org/project/frozendict/)?

Personally, I’ve been using a wrapper around `collections.namedtuple` as an underlying data structure to create frozen dictionaries when I’ve needed something like that for a project.

guidopallemans 157 days ago | | |

When you are making str -> Any dictionaries it's quite likely you're better off with dataclasses or namedtuples anyway.

pansa2 157 days ago | |

There was a previous PEP (in 2012) with the exact same title:

https://peps.python.org/pep-0416/

Also one in 2019 for a "frozenmap":

https://peps.python.org/pep-0603/

Qem 157 days ago | |

Perhaps you used the frozen dict implementation from the pip installable boltons library: https://boltons.readthedocs.io/en/latest/dictutils.html#bolt...

hiddencost 157 days ago | |

Perhaps immutabledict?

https://pypi.org/project/immutabledict/

Strilanc 157 days ago |

This is a type that I would use a lot.

For example, I often write classes that do cacheable analysis that results in a dict (e.g. the class stores a list of tiles defined by points and users want a point-to-tiles mapping for convenience). It's worth caching those transformations, e.g. using @functools.cached_property, but this introduces a risk where any caller could ruin the returned cached value by editing it. Currently, I take the safety hit (cache a dict) instead of the performance hit (make a new copy for each caller). Caching a frozendict would be a better tradeoff.

clickety_clack 157 days ago | |

Maybe you should take a look at pyrsistent. That allows you to make frozen “maps”. You can “evolve” them into new versions of the objects and it keeps the references to the unchanged keys and values under the hood so it’s fast and memory efficient.

shadowgovt 157 days ago |

Regarding the spooky-action-at-a-distance concerns of a `.freeze()` method on dict:

`.freeze()` should probably just return a frozendict instead of in-place mutating the dict, and they should be separate types. Under the hood, you'll have to build the hashtable anyway to make the frozendict; as long as you're doing that work, you may as well build an object to contain the hashtable and just have that object be separate from the dict that birthed it.

The values referenced by both the original dict and the frozendict can be the same values; no need to clone them.

emil-lp 157 days ago | |

The point is that freeze could work in constant time, whereas the copying takes linear time.

Another alternative mentioned was `move`, which would create a frozen version in constant time and clear the original dict.

shadowgovt 156 days ago | | |

Freeze can't work in constant time if it builds a hash when the dict is frozen so that the dict can be used as a key.

If all it does is set a flag that prevents modifications, that's different.

codethief 157 days ago |

Next step: Auto-inferring the correct (most narrow) TypedDict type from a frozendict. (Think `const foo = { … } as const` in TypeScript.) I miss this TS feature in Python on the daily.

mrweasel 157 days ago | |

What would that do exactly? Auto-generate a TypedDict class?

Bringing up TypedDict is sort of interesting, because it seems like a frozen dictionary could have been implemented by PEP 705, if typing.ReadOnly was enforced at runtime, and not just a hint to a static type checker.

dmurray 157 days ago | | |

Auto-generate a TypedDict type while type checking, while doing nothing at runtime.

I expect this is not a very big ask and the various typecheckers will have versions of this soon after release.

adzm 157 days ago | |

Curious what this means for typescript as well; honestly I've only reached for as const rarely.

anentropic 157 days ago | |

Agreed... but it shouldn't need a frozendict for this

IMHO TypedDict in Python are essentially broken/useless as is

What is needed is TS style structural matching, like a Protocol for dicts

virtue3 157 days ago | |

I also SUPER LOVE this feature. Especially when you make a type union of the keys for easier indexing.

tweakimp 157 days ago | |

Can you give a Python example of a use case for this?

drhagen 157 days ago |

If this gets wide enough use, they could add an optimization for code like this:

    n = 1000
    a = {}
    for i in range(n):
        a[i] = str(i)
    a = frozendict(a)  # O(n) operation can be turned to O(1)

It is relatively easy for the JIT to detect the `frozendict` constructor, the `dict` input, and the single reference immediately overwritten. Not sure if this would ever be worth it.

_flux 157 days ago | |

Wouldn't ref-counting CPython already know that a has a single reference, allowing this optimization without needing any particular smart JIT?

zbentley 157 days ago | | |

I think GP was talking about optimizing away the O(N) call on the last line. The GC will take care of removing the reference to the old (mutable) dict, but constructing a new frozendict from a mutable dict would, in the current proposal, be an O(N) shallow copy.

There are also potentially other optimizations that could be applied (not specific to dict/frozendict) to reduce the memory overhead on operations like "a = f(a)" for selected values of "f".

zahlman 157 days ago | | |

First thought: I would very much expect it to be able to do this optimization given the similar things it does for string concatenation.

But actually, I suspect it can't do this optimization simply because the name `frozendict` could be shadowed.

tracnar 157 days ago | | |

Indeed, the Tcl implementation does this so e.g. `set d [dict] ; dict set d key value` can modify d in place instead of creating a copy (since everything is immutable).

the__alchemist 157 days ago |

I wish Python could move away from types-as-mutability-determiners. Or paper over them; all could have mutable and non-mutable variants, with "to_mut()" and "to_immut()" methods. And in the same vein, explicit control over whether functions mutate a variable in place, or make a local copy. Python is my first lang, and I still am unable to consistently reason about these.

Maledictus 157 days ago | |

I love Ruby's .freeze

mwsherman 157 days ago |

I’ve found C#’s frozen dictionary to be useful: https://learn.microsoft.com/en-us/dotnet/api/system.collecti...

It’s optimized for fast reads in exchange for expensive creation.

steadyelk 157 days ago |

Agree it's about time to make this built in. Other functional packages like JAX [0] are already using the concept but they build it into their library from scratch.

[0] https://flax.readthedocs.io/en/latest/api_reference/flax.cor...

lou1306 157 days ago |

Can someone give a strong rationale for a separate built-in class? Because "it prevents any unintended modifications" is a bit weak.

If you have fixed keys, a frozen dataclass will do. If you don't, you can always start with a normal dict d, then store tuple(sorted(d.items())) to have immutability and efficient lookups (binary search), then throw away d.

vlovich123 157 days ago |

I’m curious why the dict->frozendict operation is not an O(1) operation when there’s a refcnt of 1 on the dict - that resolves the spooky action at a distance problem raised and solves the performance for the most common usage pattern (build up the dict progressively and convert to a frozendict for concurrent reads).

emil-lp 157 days ago | |

Because optimizations are not added to the PEP, but added later.

kleiba 157 days ago |

One nice side effect of a dict that's immutable is that under the hood, you could employ perfect hashing so that the your dict uses a hash function that is collision-free.

mont_tag 157 days ago |

Note that Python already offers MappingProxy that wraps a dictionary to hide mutating methods.

People hardly ever use this tool. That suggests that there isn't much of a need for a frozendict.

bvrmn 157 days ago |

It would be great to have **kwargs as frozendict by default. It could help with caching decorators to speedup key construction. For example natural key is simply `(args, kwargs)`.

DeathArrow 157 days ago |

>But dictionaries are mutable, which makes them problematic for sharing data in concurrent code.

Not really, C# has ConcurrentDictionary.

ok123456 157 days ago |

Ruby has had frozen dictionaries since version 1.0, and they are generally considered a good thing to use where possible.

xamuel 157 days ago |

This subject always seems to get bogged down in discussions about ordered vs. unordered keys, which to me seems totally irrelevant. No-one seems to mention the glaring shortcoming which is that, since dictionary keys are required to be hashable, Python has the bizarre situation where dicts cannot be dict keys, as in...

{{'foo': 'bar'}: 1, {3:4, 5:6}: 7}

...and there is no reasonable builtin way to get around this!

You may ask: "Why on earth would you ever want a dictionary with dictionaries for its keys?"

More generally, sometimes you have an array, and for whatever reason, it is convenient to use its members as keys. Sometimes, the array in question happens to be an array of dicts. Bang, suddenly it's impossible to use said array's elements as keys! I'm not sure what infuriates me more: said impossibility, or the python community's collective attitude that "that never happens or is needed, therefore no frozendict for you"

zahlman 155 days ago | |

> the glaring shortcoming which is that, since dictionary keys are required to be hashable, Python has the bizarre situation where dicts cannot be dict keys

There is nothing at all bizarre or unexpected about this. Mutable objects should not be expected to be valid keys for a hash-based mapping — because the entire point of that data structure is to look things up by a hash value that doesn't change, but mutating an object in general changes what its hash should be.

Besides which, looking things up in such a dictionary is awkward.

> More generally, sometimes you have an array, and for whatever reason, it is convenient to use its members as keys.

We call them lists, unless you're talking about e.g. Numpy arrays with a `dtype` of `object` or something. I can't think of ever being in the situation you describe, but if the point is that your keys are drawn from the list contents, you could just use the list index as a key. Or just store key-value tuples. It would help if you could point at an actual project where you encountered the problem.

kzrdude 157 days ago | |

Turning a dictionary into a tuple of tuples `((k1, v1), (k2, v2), ...)`; isn't that a reasonable way?

If you want to have hash map keys, you need to think about how to hash them and how to compare for equality, it's just that. There will be complications to that such as floats, which have a tricky notion of equality, or in Python mutable collections which don't want to be hashable.

xamuel 157 days ago | | |

That argument would apply to sets too, and yet frozenset is builtin.

whimsicalism 157 days ago |

yes! no more `MappingProxyType` hacks

drhagen 157 days ago |

Great! Now make `set` have a stable order and we're done here.

cr125rider 157 days ago | |

Aren’t sets unsorted by definition? Or do repeated accesses without modification yield different results?

ledauphin 157 days ago | | |

this is likely in reference to the fact that dicts have maintained insertion order since Python ~3.6 as property of the language. Mathematically there's no defined order to a set, and a dict is really just a set in disguise, but it's very convenient for determinism to "add" this invariant to the language.

adrian_b 157 days ago | | |

Not related to Python, but one of the possible implementations of a set, i.e. of an equivalence class on sequences, is as a sorted array (with duplicates eliminated, unless it is a multiset, where non-unique elements are allowed in the sorted array), as opposed to the unsorted array that can store an arbitrary sequence.

So sets can be viewed as implicitly sorted, which is why the order of the elements cannot be used to differentiate two sets.

Being sorted internally to enforce the equivalence between sets with elements provided in different orders does not imply anything about the existence of an operation that would retrieve elements in a desired order or which would return subsets less or greater than a threshold. When such operations are desired, an order relation must be externally defined on the set.

So a possible definition of sets and multisets is as sorted arrays with or without unicity of elements, while sequences are unsorted arrays (which may also have the constraint of unique elements). However the standard set operations do not provide external access to the internal order, which is an order between arbitrary identifiers attached to the elements of the set, which have no meaning externally.

Retr0id 157 days ago | | |

A stable order does not imply sorted order. If you convert the same set to a list twice you'll probably get the same order both times but it isn't guaranteed anywhere, and that order may change between python implementations and versions. The order may also change as the set grows or shrinks.

sltkr 157 days ago | | |

So are dictionary keys, but Python decided to make them insertion ordered (after having them be unordered just like set elements for decades). There is no fundamental reason sets couldn't have a defined order. That's what languages like JavaScript have done too.

westurner 157 days ago |

PMap and PVector, functional Python libraries,

"PEP 351 – The freeze protocol" (2005, rejected) https://peps.python.org/pep-0351/ ; IIUC the freeze protocol proposed basically:

  def freeze(obj)
    return cache.setsefault(hash(obj),
      obj.__freeze__())

/? "Existing threads re: consts and applications thereof"

I wasn't able to find a URL to this post (2021) from the python-ideas mailing list archives using a double quoted search term today; I had to use the python mailing list search engine? Did something break crawling of mailing lists? Old mailman HTML archives were very simple to crawl.. ENH: pypa: add a sitemap.xml for/of mailing list archives, forums; @pypa: ask for search engine indexing advice: "How do we make sure that the python mailing list archives will be search indexed?" (as they traditionally were)

How to find the .txt of mailing list archives posts these days?

From "[Python-ideas] Re: Introduce constant variables in Python" (2021) https://mail.python.org/archives/list/python-ideas@python.or... :

- pyrsistent: PMap, PVector

I'm out of time for this; (reformatting this for HN so that URLs will be auto-linkified but newlines won't be eliminated) here's the full email as .txt, the mailing list archive has a hyperlinkified version with newlines preserved. GH Markdown and CommonMark Markdown also preserve newlines and auto-linkify:

  From: [@westurner]
  Date: Thu, Jun 17, 2021, 10:43 AM
  Subject: Re: [Python-ideas] Re: Introduce constant variables in Python
  Cc: python-ideas <python-ideas@python.org>
  
  
  On Mon, May 24, 2021 at 5:43 PM Chris Angelico <rosuav@gmail.com> wrote:
  
  Requiring that a name not be rebound is well-defined and testable.
  Requiring that an object not change is either trivial (in the case of,
  say, an integer) or virtually impossible (in the case of most
  objects).
  
  What would be the advantage of such a declaration?
  
  ChrisA
  
  ## Existing threads re: consts and applications thereof
  
   So, `/? from:me pyrsistent` I found a few results:
  - "[Python-Dev] Challenge: Please break this! (a.k.a restricted mode revisited)" 2016-04
    https://mail.python.org/pipermail/python-dev/2016-April/143958.html
    - ~Sandboxing python within python is nontrivial to impossible; consts might help a bit
      - https://mail.python.org/pipermail/python-dev/2016-April/143958.html
  
  - "Proposal to add const-ness type hints to PEP-484"
    https://mail.python.org/archives/list/python-ideas@python.org/thread/OVPF5I6IOVF6GOJQRH5UGCCU3R7PQHUF/
    - https://github.com/python/typing/issues/242
      - "Final names and attributes" https://github.com/python/mypy/pull/5522
         This is where `typing.Final` comes from.
  
  - "[Python-ideas] "Immutable Builder" Pattern and Operator"
    https://mail.python.org/pipermail/python-ideas/2017-January/044374.html
    - [pyrsistent] and "fn.py [do] immutables:
      https://github.com/kachayev/fn.py/blob/master/README.rst#persistent-data-structures "
  
  - "[Python-ideas] Add recordlcass to collections module"
    https://groups.google.com/g/python-ideas/c/9crHfcCBgYs/m/6_EEaWJAAgAJ
    - ORMs (e.g. Django, SQLAlchemy) require "dirty state" checking to know which object attributes have changed and need an SQL statement to be executed to synchronize the state; this is relevant because when we're asking for mutable namedtuple we're often trying to do exactly this pattern.
  
  - "[Python-ideas] Suggestions: dict.flow_update and dict.__add__"
    https://www.google.com/search?q=%22%5BPython-ideas%5D+Suggestions%3A+dict.flow_update+and+dict.__add__%22
  
    > dicttoolz has functions for working with these objects; including dicttoolz.merge (which returns a reference to the merged dicts but does not mutate the arguments passed).
    > 
    > https://toolz.readthedocs.io/en/latest/api.html#dicttoolz
    > https://toolz.readthedocs.io/en/latest/api.html#toolz.dicttoolz.merge
    > 
    > pyrsistent has a PRecord class with invariants and type checking that precedes dataclasses. pyrsistent also has 'freeze' and 'thaw' functions for immutability. PRecord extends PMap, which implements __add__ as self.update(arg) (which does not mutate self)
  https://github.com/tobgu/pyrsistent/blob/master/README.rst#precord
    > 
    > https://github.com/tobgu/pyrsistent/blob/master/pyrsistent/_pmap.py
  
  - "[Python-ideas] How to prevent shared memory from being corrupted ?"
    https://www.google.com/search?q=%22How+to+prevent+shared+memory+from+being+corrupted+%3F%22
    > PyArrow Plasma object ids, "sealing" makes an object immutable, pyristent
    > 
    > https://arrow.apache.org/docs/python/plasma.html#object-ids
    > https://arrow.apache.org/docs/python/plasma.html#creating-an-object-buffer
  
    > > Objects are created in Plasma in two stages. First, they are created, which allocates a buffer for the object. At this point, the client can write to the buffer and construct the object within the allocated buffer. [...]
  
  - [Python-ideas] Experimenting with dict performance, and an immutable dict
    https://mail.python.org/archives/list/python-ideas@python.org/message/DNBGUJHDH4UTPSETMFFWMJHNXQXIWX4I/
  
    > https://pyrsistent.readthedocs.io/en/latest/intro.html#pyrsistent :
    > 
    >> Pyrsistent is a number of persistent collections (by some referred to as functional data structures). Persistent in the sense that they are immutable.
    >>
    >> All methods on a data structure that would normally mutate it instead return a new copy of the structure containing the requested updates. The original structure is left untouched.
    >>
    >> This will simplify the reasoning about what a program does since no hidden side effects ever can take place to these data structures. You can rest assured that the object you hold a reference to will remain the same throughout its lifetime and need not worry that somewhere five stack levels below you in the darkest corner of your application someone has decided to remove that element that you expected to be there.
    >>
    >> Pyrsistent is influenced by persistent data structures such as those found in the standard library of Clojure. The data structures are designed to share common elements through path copying. It aims at taking these concepts and make them as pythonic as possible so that they can be easily integrated into any python program without hassle.
  
    
  
  > What would be the advantage of such a declaration?
  
  Constants don't need to be locked or unlocked; which is advantageous for parallelism and reasoning about program correctness.
  True consts (wherein everything referred to in that object is 'frozen' and immutable or at least only modifiable with e.g. copy-on-write)
  wouldn't require locks,
  which would be post-GIL advantageous.
  
  You could do consts by never releasing a threading.Lock (or similar):
  - https://docs.python.org/3/library/asyncio-sync.html#locks
  - https://docs.python.org/3/library/threading.html#lock-objects
  - This from 
  https://docs.python.org/2/library/sets.html?highlight=immutable#immutable-transforms re ImmutableSet/FrozenSet
  is not present in the python 3 docs: 
  https://docs.python.org/3/library/stdtypes.html#set-types-set-frozenset 
  
  Though - even if Python enforced normal consts in the language - all of the other code objects would still be mutable, so you still have the impossibility of sandboxing python.
  
  Functional and contracts coding styles rely upon invariance; 
  which can be accomplished with various third-party packages that enforce const-ness throughout what may be an object tree behind that reference that would otherwise need to be copy.deepcopy()'d.
  
  ## pyrsistent
  Src: https://github.com/tobgu/pyrsistent
  
  > - PVector, similar to a python list
  > - PMap, similar to dict
  > - PSet, similar to set
  > - PRecord, a PMap on steroids with fixed fields, optional type and invariant checking and much more
  > - PClass, a Python class fixed fields, optional type and invariant checking and much more
  > - Checked collections, PVector, PMap and PSet with optional type and invariance checks and more
  > - PBag, similar to collections.Counter
  > - PList, a classic singly linked list
  > - PDeque, similar to collections.deque
  > - Immutable object type (immutable) built on the named tuple
  > -   freeze and thaw functions to convert between python standard collections and pyrsistent collections.
  > - Flexible transformations of arbitrarily complex structures built from PMaps and PVectors.
  
  
  ## icontract
  Src: https://github.com/Parquery/icontract
  
  > icontract provides design-by-contract to Python3 with informative violation messages and inheritance.
  >
  > It also gives a base for a flourishing of a wider ecosystem:
  > 
  > - A linter pyicontract-lint,
  > - A sphinx plug-in sphinx-icontract,
  > - A tool icontract-hypothesis for automated testing and ghostwriting test files which infers Hypothesis strategies based on the contracts,
  together with IDE integrations such as icontract-hypothesis-vim, icontract-hypothesis-pycharm, and icontract-hypothesis-vscode,
  > - Directly integrated into CrossHair, a tool for automatic verification of Python programs,
  together with IDE integrations such as crosshair-pycharm and crosshair-vscode, and
  > - An integration with FastAPI through fastapi-icontract to enforce contracts on your HTTP API and display them in OpenAPI 3 schema and Swagger UI.

https://en.wikipedia.org/wiki/Design_by_contract

https://en.wikipedia.org/wiki/Invariant_(mathematics)#Invari... [ https://en.wikipedia.org/wiki/Class_invariant ]

> What is the difference between "invariant" and "constant" and "final"?

semiinfinitely 157 days ago |

its called dataclass

malkia 157 days ago |

Welcome to starlark :)