gh-77753: Add hash examples for set/dict - #92549
Conversation
Sorry, something went wrong.
| >>> d = {False: 'false'} | ||
| >>> d[0] = 0 | ||
| >>> d | ||
| {False: 0} |
There was a problem hiding this comment.
| {False: 0} | |
| {False: 0} | |
| >>> {1: 'a', True: 'b'} | |
| {1: 'b'} |
Sorry, something went wrong.
| Note that as long as the objects have the same hash values, they are seen as | ||
| equivalent by sets:: | ||
|
|
||
| >>> {0, True, False, 1} |
There was a problem hiding this comment.
Could you please add some of the types Terry listed in the issue too? I was frankly surprised at them (especially the Decimal and Fractions with the same hash value).
Maybe this would help:
| >>> {0, True, False, 1} | |
| >>> {0, True, False, 1, 0.0, 1.0} |
Sorry, something went wrong.
| Note that as long as the objects have the same hash values, they are seen as | ||
| equivalent by sets:: |
There was a problem hiding this comment.
It is not true.
>>> hash(1) == hash(2**61)
True
>>> {1, 2**61}
{1, 2305843009213693952}
Sorry, something went wrong.
| This also means if two keys have the same hash values, they will map to the same | ||
| entry in the dictionary:: |
There was a problem hiding this comment.
Same as above.
Sorry, something went wrong.
|
Besides making two flat-out incorrect statements, I also don't think this is very useful. The examples are cute but aren't relevant to most users of dicts and sets. Also, it isn't the sets or dicts themselves that give rise to these examples; instead, it is that numbers were intentionally designed to support cross-type equality comparisons and to follow the rule that equal values should have the same hashes. This is documented elsewhere. Thanks for the suggestion, but I'm marking it is closed as a low quality edit. |
Sorry, something went wrong.
|
@rhettinger I understand that the examples on hashing isn't as useful, but does the dict example about same keys but different values not warrant at least something? There seems to be twoish open issues on this, and I don't think the behavior's exactly intuitive. If it is this behavior that's documented somewhere, then I reckon that some sort of link or reference would help. When you mark it closed due to low quality, is it due to examples not thorough enough, wording too short, or something else? I would like to improve the docs so knowing what exactly can be changed other than the incorrectness would be helpful to me. |
Sorry, something went wrong.
#77753
https://docs.python.org/3/library/stdtypes.html#set-types-set-frozenset
Depended by #91499