distinctUntilChanged should set prev value with keySelector if available even at the first time

[preda7or]

Bug Report

Current Behavior
distinctUntilChanged on master seem not to set the prev value correctly the first time.

Expected behavior
the prev value should be set with keySelector if available even at the first time.

Reproduction

** Please provide repro code does not involves framework, or other setups to address this bug is specific to rxjs core **
f467aa2?branch=f467aa25da69859d1540413c6c2d5a289f2c88d2&diff=split

Environment

• Runtime: -
• RxJS version: master branch

Possible Solution
see above

Additional context/Screenshots

[cartant]

If this issue relates to a bug effected by the current implementation, it needs to include a code snippet that effects the bug. TBH, I don't understand what's being said here.

[preda7or]

So the commit I linked does not count as code snippet?

[cartant]

TBH, I don't understand what's being said here.

[preda7or]

What I was trying to say, thinking that someone will check the linked commit too, that the refactored code of distinctUntilChanged operator looks incorrect to me:

When 'first' is true (so the first time), the 'prev' variable is set to be 'value'. In every subsequent run (when 'first' is false) 'prev' is set via a ternary operator that uses the 'keySelector' function if it is defined.
So if I have a keySelector set, the first time distinctUntilChanged operator is called the value of 'prev' will be set WITHOUT the keySelector, then in the next call of distinctUntilChanged this 'prev' value will be compared with the 'value' processed WITH the keySelector. So even if the 'value' is the same, this comparison will return false.

[cartant]

... the refactored code of distinctUntilChanged operator looks incorrect to me.

That's why I want to see a snippet that shows it won't behave has expected - i.e. I want to see a reproduction of a bug. The code looks correct to me and I don't want to spend my time investigating something that looks correct.

[preda7or]

I don't want to spend my time investigating something that looks correct

Fair enough, no one wants to work.

So I created a PR, that contains a new test case and the fixed operator. Hopefully this will help, I cannot be more explicit then this: #6013

[cartant]

Okay, I see what you mean. I presumed you'd commented on a commit that had been made in this repo.