gh-111545: Add Py_HashDouble() function#112449
Conversation
|
@encukou @mdickinson @serhiy-storchaka: Would you mind to review this change? It's a new API for PyHash_Double() function: other APIs were discussed in PR #112095. |
Sorry, something went wrong.
ae3843d to
e69656a
Compare
November 30, 2023 22:31
|
I renamed the function from |
Sorry, something went wrong.
mdickinson
left a comment
There was a problem hiding this comment.
The API looks fine to me. I'm wondering whether we want to consider switching the meaning of the return value, so that 0 is used for non-NaNs and 1 for NaNs; I want to think of the return value as essentially an is_nan check, and from that perspective the current API has an extra negation to get my brain around.
But either way is probably fine (so long as it's documented).
Sorry, something went wrong.
e69656a to
8e29a02
Compare
December 3, 2023 11:10
|
I rebased the PR on the main branch.
Ok, I modified the API to return 1 if the float is not-a-number (NaN), and return 0 otherwise. |
Sorry, something went wrong.
|
In similar API, 1 means that you're getting a meaningful IMO, that's a good convention to establish (here, with |
Sorry, something went wrong.
Fair enough. That convention does feel somewhat in conflict with the common "zero-return = success", "non-zero-return = failure" API that we have elsewhere, but that's a wider issue. |
Sorry, something went wrong.
Agreed. |
Sorry, something went wrong.
If I'm reading the room correctly, we're reserving only |
Sorry, something went wrong.
49bae56 to
5be68f0
Compare
December 4, 2023 13:50
|
I reverted my change to move back to the previous API:
@mdickinson @encukou: Would you mind to review/approve formally the PR? |
Sorry, something went wrong.
encukou
left a comment
There was a problem hiding this comment.
Looks good, thanks!
Other WG members might disagree, but I don't see anything too controversial any more.
Sorry, something went wrong.
I created capi-workgroup/decisions#2 for the C API Working Group. |
Sorry, something went wrong.
5be68f0 to
155764a
Compare
December 6, 2023 14:37
|
I merged my Py_HashPointer() PR. I rebased this PR on top on it. |
Sorry, something went wrong.
|
I will update the PR to address the reviews, once the API will be approved by the C API Working Group. |
Sorry, something went wrong.
* Add again the private _PyHASH_NAN constant. * Add tests: Modules/_testcapi/hash.c and Lib/test/test_capi/test_hash.py.
encukou
left a comment
There was a problem hiding this comment.
LGTM, thank you!
Sorry, something went wrong.
|
I updated What's New in Python 3.13 to suggest a recipe replacing usage of the private |
Sorry, something went wrong.
|
I would prefer a simpler API that does not treat NaN as special at all and leave this to the user: Even if it is slightly slower. But it requires less mental effort to understand and use the API. |
Sorry, something went wrong.
What should an user pass in as |
Sorry, something went wrong.
@serhiy-storchaka proposes the API: I would be fine with this API. It was more or less proposed earlier. But it was said that having to call In terms of API,
|
Sorry, something went wrong.
|
I wrote PR #113115 which implements |
Sorry, something went wrong.
|
@serhiy-storchaka @mdickinson: So which API do you prefer?
|
Sorry, something went wrong.
|
I prefer your initial interface Unless there is really large performance advantage in using the second variant. I do not know how large should it be to justify more complex interface. 10% is not large. |
Sorry, something went wrong.
Ok.
If prefer to have a deterministic behavior and always return the same hash value (0) if value is NaN. There are legit use cases to treat NaN as hash value 0. See my comment. |
Sorry, something went wrong.
Ditto. I don't see a need for anything more complicated than this, and it feels wrong to me to allow minor performance differences to drive API design.
This sounds good to me. |
Sorry, something went wrong.
Ok. I proposed to change C API Working Group decision on this API: capi-workgroup/decisions#2 (comment) |
Sorry, something went wrong.
It creates a vulnerability. CPython itself never calls this API for a NaN. If a third-party code calls it for a NaN value, it will allow to easily create non-equal objects with the same hash. |
Sorry, something went wrong.
|
The simpler API,
|
Sorry, something went wrong.
Can this concern be resolved with better documentation? Explain in which case treating all NaN "as equal" can be an issue, or suggest a solution such as @serhiy-storchaka's recipe #112449 (comment) ? I'm not convinced that using the same hash value (0) for all NaN numbers is a big deal, since Python was doing that until 3.9. The hash value is just an optimization to avoid the slower comparison operator, it's more about performance than about correctness. |
Sorry, something went wrong.
IMO, no. When doing a review, you don't check the docs of all functions you see. Just the surprising ones. |
Sorry, something went wrong.
|
It is a big deal. Try to create a dict with 10000 or 100000 of NaNs. d = {float('nan'): i for i in range(100000)} |
Sorry, something went wrong.
Aha, I see: it's way faster in Python 3.10 where hash(nan) is not always 0. But the proposed API is for For me, |
Sorry, something went wrong.
I'm talking about |
Sorry, something went wrong.
I'm also talking about the documentation. Users won't read the docs. If the function has surprising behaviour, adding a hint that you should look at the docs goes a long way. |
Sorry, something went wrong.
|
I created PR #112095 more than 1 month ago. I spent time to run benchmark, implement different APIs, try to collect feedback on each API, and discuss in length advantages and disadvantages of each API. Sadly, we failed to reach a consensus on the API. Now another API is being discussed. The API looks simple to me, I didn't expect to spend more than one month on a single function. I need to take a break from that topic. I don't have the energy to dig into these discussions. I prefer to close the PR for now. |
Sorry, something went wrong.
_Py_HashDoublepublic again as "unstable" API #111545📚 Documentation preview 📚: https://cpython-previews--112449.org.readthedocs.build/