bug-gnulib
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

gnulib-tool.py: Make GLModule's __eq__ and __hash__ method agree.


From: Collin Funk
Subject: gnulib-tool.py: Make GLModule's __eq__ and __hash__ method agree.
Date: Tue, 16 Apr 2024 12:50:04 -0700
User-agent: Mozilla Thunderbird

>From the documentation of object.__hash__(self) [1]:

     The only required property is that objects which compare equal
     have the same hash value; it is advised to mix together the hash
     values of the components of the object that also play a part in
     comparison of objects by packing them into a tuple and hashing
     the tuple.

Right now we use the following to compute the hash in GLModule:

    result = hash(self.name) ^ hash(self.patched)

The way that the documentation recommends writing this:

    result = hash((self.name, self.patched))

But I believe the 'patched' boolean should not be involved in
computing the hash. This would make __hash__ align with the definition
of __eq__. This also seems to better match the way dependencies are
tracked using the following string as a key:

    module1---module2

Since __hash__ is also used for sets using the 'patched' would allow a
set with an unpatched variant and patched variant in it, which I
assume would cause issues. In practice, I think you would have to
create & remove files to make the lookups in GLFileSystem return
these.

Since __hash__ is used a lot from sets I thought it would be
interesting to look at the profiler output as well. These are all from
test-create-testdir-4.sh.

Using hash(self.name) ^ hash(self.patched) as we have now:

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
   504489    0.183    0.000    0.294    0.000 GLModuleSystem.py:256(__hash__)
  1008978    0.112    0.000    0.112    0.000 {built-in method builtins.hash}

Using hash((self.name, self.patched)) as recommended:

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
   504489    0.119    0.000    0.184    0.000 GLModuleSystem.py:256(__hash__)
   504489    0.066    0.000    0.066    0.000 {built-in method builtins.hash}

Using hash(self.name) which as in the patch I attached:

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
   504489    0.102    0.000    0.166    0.000 GLModuleSystem.py:256(__hash__)
   504489    0.064    0.000    0.064    0.000 {built-in method builtins.hash}

[1] https://docs.python.org/3/reference/datamodel.html#object.__hash__

Collin

Attachment: 0001-gnulib-tool.py-Make-GLModule-s-__eq__-and-__hash__-m.patch
Description: Text Data


reply via email to

[Prev in Thread] Current Thread [Next in Thread]