Lint suggests "not (isNull x)" over "x <> null"

Description

The linter suggests replacing a null check (x <> null) with the more complicated not (isNull x). I’m not coming up with any cases where I think a null check would be better as not (isNull x).

Repro steps

Please provide the steps required to reproduce the problem

1. Add a <> null check to your F# code

2. View linter error

Expected behavior

The linter doesn’t warn on <> null checks.

Actual behavior

The linter warns on <> null checks.

Known workarounds

Disable the linter.

Related information

• Operating system
• Branch
• .NET Runtime, CoreCLR or Mono Version
• Performance information, links to performance testing scripts

[asik]

Null checks are difficult. And not using <> null is a good suggestion: https://latkin.org/blog/2015/05/18/null-checking-considerations-in-f-its-harder-than-you-think/

I wonder if this is still true today. Lots of work went into optimization in both F# and the JIT since then. I tried it quickly on SharpLab (link) and the generated code was properly inlined.

Well, nevermind, it`s still bad.

[jgardella]

@abelbraaksma Yes it is, but we changed from using attributes to using comments for that. But you cannot disable a specific hint, you can only disable all Hints. In this case you could add something like // fsharplint:disable Hints.

I had suggested removing the hint from the config itself as then you can disable just that hint, and it sounded like that’s what was wanted. But I agree with you that this is useful hint and probably best to keep it on by default.