Enum allows numeric keys with leading 0s

[laurenz-rausche]

🔎 Search Terms

"enum leading 0", "enum numeric"

🕗 Version & Regression Information

• This is the behavior in every version I tried, and I reviewed the FAQ for entries about "enums"

⏯ Playground Link

https://www.typescriptlang.org/play/?ts=5.5.0-dev.20240504#code/KYOwrgtgBAKsDOAXAouCBvAUFKAiA+gAy5QC8eAJgPYJQDuVATgNZQCG8UAlolPABZUwAGwq5seYmTxRqtEFV4MW7Tjz6CRFKAsS4ANBNyEp5XMuacARmF4ChonYoNHCARhJn6TS1Bt3NR10XAF9MTDgkVEgAOiIIhBQ0GMIEqOSTNKTY9yA

💻 Code

enum TestEnum{
  "_0" = "does work as it should"
  "0" = " does not work as it should not",
  "00" = "works but should not",
  "01" = " works but should not",
}

TestEnum._0
TestEnum.0
TestEnum.00
TestEnum.01

🙁 Actual behavior

An enum can be set with a numeric key when a leading 0 is added.

🙂 Expected behavior

The enum keys "00", "01", ... should not be allowed to be set, as they are unvalid to write as TestEnum.00, TestEnum.01, ...

Additional information about the issue

No response

[xiBread]

Bracket notation is still valid: TestEnum["00"]

[laurenz-rausche]

Yeah, but in that sense TestEnum["0"] would also be valid.

[fatcerberus]

It's not about whether or not you can write TestEnum.Foo (e.g. "Foo Bar" = "works" is also legal) but whether the name of a member could potentially clash with a reverse mapping, so keys that overlap with the canonical string representation of a number are intentionally not allowed.

[jcalz]

This isn't a bug: all of the following are intentionally allowed:

enum TestEnum {
  "1e2",
  "3.40",
  ".567",
  "88888888888888888"
}

It's for the reason @fatcerberus said; while those can be parsed as numbers, they are not how those numbers would be represented when converted to strings, so they cannot conflict with reverse mappings. If there were no reverse mappings, all numeric keys would be allowed.

[laurenz-rausche]

Oh okay! Thank you for the information, and sorry for the wrong assumption and the disturbance :D