Fix hashing for custom eq objects

[OneRaynyDay]

Summary

Thanks for making attrs btw! It's an amazing library. In the documentation for attr.ib (sorry, still speaking in 2021 attr here), we see:

hash (Optional[bool]) – Include this attribute in the generated hash method. If None (default), mirror eq’s value. This is the correct behavior according the Python spec. Setting this value to anything else than None is discouraged.

Currently this does not happen - if we passed in a callable to transform the field's value, we would see _x_key(self.x) in the generated eq code, but we don't see the same thing happening for the generated hash function.

Currently, I'd like to compare equality of a non-hashable object. I was surprised to see that the hash function did not respect the eq override and get it for free (which is an awesome feature from attr!).

Pull Request Check List

• Added tests for changed code.
  Our CI fails if coverage is not 100%.
• New features have been added to our Hypothesis testing strategy.
• Changes or additions to public APIs are reflected in our type stubs (files ending in .pyi).
  • ...and used in the stub test file tests/typing_example.py.
  • If they've been added to attr/__init__.pyi, they've also been re-imported in attrs/__init__.pyi.
• Updated documentation for changed code.
  • New functions/classes have to be added to docs/api.rst by hand.
  • Changes to the signature of @attr.s() have to be added by hand too.
  • Changed/added classes/methods/functions have appropriate versionadded, versionchanged, or deprecated directives.
    Find the appropriate next version in our __init__.py file.
• Documentation in .rst files is written using semantic newlines.
• Changes (and possible deprecations) have news fragments in changelog.d.

As the change is only meant to make the code be compliant with the docs, I have not updated any existing documentation. If this is worth adding into the changelog let me know and I'll add it asap. Thank you!

[hynek]

This looks great, thanks! Always nice to see users caring and fixing subtle edge cases. :)

This is definitely worth a changelog entry, but I wouldn't tie it to the doc entry.

Just something along the lines of "If an eq key is defined, it is also used before hashing the attribute." just in more understandable. 😇

Thanks!

P.S. also looks like coverage is failing now 😇

[OneRaynyDay]

Hi @hynek , thanks for getting back to me! I've not used this coverage tool, and I was wondering what I needed to do to bring the coverage back up to 100%. I updated the changelog.d with the update :)

[hynek]

Since I don't have push permissions against your branch, please apply the following patch and try again:

diff --git src/attr/_make.py src/attr/_make.py
index 962e560..86ced77 100644
--- src/attr/_make.py
+++ src/attr/_make.py
@@ -325,13 +325,11 @@ def _compile_and_eval(script, globs, locs=None, filename=""):
     eval(bytecode, globs, locs)
 
 
-def _make_method(name, script, filename, globs=None):
+def _make_method(name, script, filename, globs):
     """
     Create the method with the script given and return the method object.
     """
     locs = {}
-    if globs is None:
-        globs = {}
 
     # In order of debuggers like PDB being able to step through the code,
     # we add a fake linecache entry.
@@ -1955,7 +1953,7 @@ if HAS_F_STRINGS:
         ]
 
         return _make_method(
-            "__repr__", "\n".join(lines), unique_filename, globs=globs
+            "__repr__", "\n".join(lines), unique_filename, globs
         )
 
 else:

[hynek]

thanks!