Equate non-native functions that have the same code. #12
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
benjamn
force-pushed
the
equate-same-code-functions
branch
from
ec4a2a8 to
bc65262
Compare
benjamn
added a commit
to apollographql/apollo-client
that referenced
this pull request
Jul 16, 2020
benjamn
added a commit
to apollographql/apollo-client
that referenced
this pull request
Jul 20, 2020
benjamn
added a commit
that referenced
this pull request
Oct 1, 2020
Previously, an object `a` with a key `k` whose value is undefined (that is, `a[k] === void 0`) would not be considered deeply equal to another object `b` that is missing that key, but is otherwise deeply equal to the first object, even though `a[k] === b[k]`. This commit makes these two different kinds of undefined-ness equivalent, by efficiently ignoring undefined-valued keys when comparing objects. This allows more objects than before to be considered equivalent, but the equal(a, b) function still obeys the rules of equivalence relations: reflexivity, symmetry, and transitivity. This change is similar in spirit to the way we relaxed the strict === requirement for function objects in #12: more functions can be considered equivalent now, but the newly equivalent functions still obey the rules of equality, and thus the change is logically valid, a decision we are free to make but not forced to make. The additional leniency will be useful in situations where (for example) multiple options objects are combined by application code and then passed to an API, perhaps using ...spread syntax or Object.assign, and the API needs to decide if the new options are the same as previous options. This change allows the application code to avoid worrying about the difference between undefined and missing object properties, a discipline that can unnecessarily complicate the way the options need to be combined.
benjamn
added a commit
that referenced
this pull request
Oct 1, 2020
Previously, an object `a` with a key `k` whose value is undefined (that is, `a[k] === void 0`) would not be considered deeply equal to another object `b` that is missing that key, but is otherwise deeply equal to the first object, even though `a[k] === b[k]`. This commit makes these two different kinds of undefined-ness equivalent, by efficiently ignoring undefined-valued keys when comparing objects. This allows more objects than before to be considered equivalent, but the equal(a, b) function still obeys the rules of equivalence relations: reflexivity, symmetry, and transitivity. This change is similar in spirit to the way we relaxed the strict === requirement for function objects in #12: more functions can be considered equivalent now, but the newly equivalent functions still obey the rules of equality, and thus the change is logically valid, a decision we are free to make but not forced to make. The additional leniency will be useful in situations where (for example) multiple options objects are combined by application code and then passed to an API, perhaps using ...spread syntax or Object.assign, and the API needs to decide if the new options are the same as previous options. This change allows the application code to avoid worrying about the difference between undefined and missing object properties, a discipline that can unnecessarily complicate the way the options need to be combined.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Previously, function objects were regarded as equal by
@wry/equalityonly if they were===to each other. This policy was overly restrictive in many common situations, such as when comparing functions that are repeatedly passed as fresh function expressions within objects that are otherwise deeply equal.Thanks to this PR, the
@wry/equalitypackage now considers non-native functions equal if they are===(as before) or if they have the same code according toFunction.prototype.toString. For native functions, we must check===because their code is censored, e.g.Object.assign.toString() === "function assign() { [native code] }".Note that this behavior is not entirely sound, since
!==function objects with the same code can behave differently depending on their closure scope. However, any function can behave differently depending on the values of its input arguments (includingthis) and its calling context (including its closure scope), even though the function object is obviously===to itself; and it is entirely possible for functions that are not===to behave exactly the same under all conceivable circumstances.Because none of these factors are statically decidable in JavaScript (or any Turing-complete programming language, probably?), JS function equality is simply not well-defined. This ambiguity allows us to consider the best possible heuristic among various imperfect options, and equating non-native functions that have the same code has enormous practical benefits (see apollographql/apollo-client#6588, for example).
I considered making this behavior configurable, until I realized that strict
===equality was never an absolute guarantee of behavioral equivalence, anyway. Ambiguities can be liberating!