Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We鈥檒l occasionally send you account related emails.

Already on GitHub? Sign in to your account

gh-111545: Add Py_HashDouble() function #113115

Closed
wants to merge 2 commits into from

Conversation

vstinner
Copy link
Member

@vstinner vstinner commented Dec 14, 2023

Add tests: Modules/_testcapi/hash.c and
Lib/test/test_capi/test_hash.py.


馃摎 Documentation preview 馃摎: https://cpython-previews--113115.org.readthedocs.build/

Add tests: Modules/_testcapi/hash.c and
Lib/test/test_capi/test_hash.py.
@@ -84,17 +84,20 @@ static Py_ssize_t hashstats[Py_HASH_STATS_MAX + 1] = {0};
*/

Py_hash_t
_Py_HashDouble(PyObject *inst, double v)
Py_HashDouble(double v)
{
int e, sign;
double m;
Py_uhash_t x, y;

if (!Py_IS_FINITE(v)) {
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What if remove this and keep only Py_IS_INFINITY(v) check?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

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.

With the following change, only check for, Py_HashDouble() hangs (fail to exit the loop) if value is NaN.

diff --git a/Python/pyhash.c b/Python/pyhash.c
index f64edde4043..23aa2dac7cc 100644
--- a/Python/pyhash.c
+++ b/Python/pyhash.c
@@ -90,14 +90,8 @@ Py_HashDouble(double v)
     double m;
     Py_uhash_t x, y;
 
-    if (!Py_IS_FINITE(v)) {
-        if (Py_IS_INFINITY(v)) {
-            return (v > 0 ? _PyHASH_INF : -_PyHASH_INF);
-        }
-        else {
-            assert(Py_IS_NAN(v));
-            return 0;
-        }
+    if (Py_IS_INFINITY(v)) {
+        return (v > 0 ? _PyHASH_INF : -_PyHASH_INF);
     }
 
     m = frexp(v, &e);

With the following change, Py_HashDouble() returns -_PyHASH_INF if value is NaN, since NaN > 0 is false:

diff --git a/Python/pyhash.c b/Python/pyhash.c
index f64edde4043..a853d6dad99 100644
--- a/Python/pyhash.c
+++ b/Python/pyhash.c
@@ -91,13 +91,8 @@ Py_HashDouble(double v)
     Py_uhash_t x, y;
 
     if (!Py_IS_FINITE(v)) {
-        if (Py_IS_INFINITY(v)) {
-            return (v > 0 ? _PyHASH_INF : -_PyHASH_INF);
-        }
-        else {
-            assert(Py_IS_NAN(v));
-            return 0;
-        }
+        // v can be NaN
+        return (v > 0 ? _PyHASH_INF : -_PyHASH_INF);
     }
 
     m = frexp(v, &e);

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What if use Py_IS_INFINITY() instead of !Py_IS_FINITE()?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

My first attempt (first patch in my comment) leads to a hang if you pass NaN.

Why do you want to avoid !Py_IS_FINITE + Py_IS_INFINITY check? Are you worried about performance?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Recipe of What's New in Python 3.13:

     Py_hash_t hash_double(PyObject *obj, double value)
     {
         if (!Py_IS_NAN(value)) {
             return Py_HashDouble(value);
         }
         else {
             return Py_HashPointer(obj);
         }
     }

Using this recipe and the current implementation, there are 3 code paths:

  • NaN: 1 test (Py_IS_NAN()), hash_double() calls Py_HashPointer().
  • infinity: 3 tests (!Py_IS_NAN(), !Py_IS_FINITE(), Py_IS_INFINITY()), return (v > 0 ? _PyHASH_INF : -_PyHASH_INF).
  • finite: 2 tests (!Py_IS_NAN(), Py_IS_FINITE()), the loop.

I don't think that it's a big deal to add 1 or 2 tests per float point number. I care more about the API, having a deterministic behavior for the 3 cases.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I want to avoid any promises about NaN. It should be recommended to not use this function for NaN.

Comment on lines +55 to +66
* If *value* is positive infinity, return :data:`sys.hash_info.inf
<sys.hash_info>`.
* If *value* is negative infinity, return :data:`-sys.hash_info.inf
<sys.hash_info>`.
* If *value* is not-a-number (NaN), return :data:`sys.hash_info.nan
<sys.hash_info>` (``0``).
* Otherwise, return the hash value of the finite *value* number.

.. note::
Return the hash value ``0`` for the floating point numbers ``-0.0`` and
``+0.0``, and for not-a-number (NaN). ``Py_IS_NAN(value)`` can be used to
check if *value* is not-a-number.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It exposes too much implementation details why already exposed in different place. Why not simply say that it is equivalent to hash() of Python float object if it is not a NaN? And if it is a NaN, you should use other value to avoid collisions.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Are you talking about the note, or describing the 3 cases and return values? I can just remove the note. My idea is to suggest using Py_IS_NAN() to treate NaN differently. But I'm not sure which implementation to suggest.

@zooba says that if you have a Python object, just call PyObject_Hash(obj) on it 馃榿

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

About describing all 3 cases. It should already be described in other place (documentation for sys.hash_info or float or hash()), and if it is not described in details, than it is not necessary for users. You should only document that for non-NaN values it returns the same result as for hash() for Python float object.

@vstinner
Copy link
Member Author

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.

@vstinner vstinner closed this Dec 20, 2023
@vstinner vstinner deleted the hash_double4 branch December 20, 2023 11:02
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

2 participants