Inconsistent behavior of reportUnnecessaryComparison

[Zzzen]

a = 1
b = 2
_1 = a == 2

c = True
d = False
_2 = c == d
_22 = c != d

def f():
  a = True
  b = True
  _3 = a == b
  _4 = a != b

I would expect _1/_2/_22/_3/_4 to be prompted for errors because the result of comparisons can be statically determined. But actually only _1/_2/_22 are prompted.

playground link: https://zzzen.github.io/pyright-ast-viewer/#code/FDCGAIF5wRmAjK4BMwD6MkUtVwDGSAKgE4CuApsACZIBioANgM5VrJKE7jUjUUAzcAIAUASgBcwcOGzhSlaeETQFVGWgDMWKNHhK0AFh0BCPcCA

ref:

reportUnnecessaryComparison [boolean or string, optional]: Generate or suppress diagnostics for == or != comparisons or other conditional expressions that are statically determined to always evaluate to False or True. Such comparisons are sometimes indicative of a programming error. The default value for this setting is "none".

[erictraut]

Yes, this is by design. The reportUnnecessaryComparison check is selective in what types of errors it reports. We found that there are too many legitimate cases where always-true checks are used in typical Python code. This often happens, for example, in if/elif/else chains used to exhaustively check enum values. For this reason, we don't report errors for cases like Literal[True] == Literal[True]. It generates too much noise. The purpose of this check is to highlight uses that are likely to be bugs.