Rule to recommend if-else construct instead of match clause

[webwarrior-ws]

It improves code readability.

Fixes: #545

Supersedes #588

[knocte]

From #586:

btw what do you think about renaming the rule to RecommendIfElseConstructOverMatch?

[webwarrior-ws]

From #586:

btw what do you think about renaming the rule to RecommendIfElseConstructOverMatch?

I already renamed it in one of the commits.

Also it supersedes #588, not #586. Fixed PR description.

[knocte]

@webwarrior-ws this cannot be written as if foo = Baz.Bar?

[webwarrior-ws]

Actually it can. Though in this situation pattern matching seems more natural choice.
So should I remove this case from exceptions to the rule?

[knocte]

yes let's remove it because I want to see what's the impact on SelfCheck; keep your change around in case we put it back

[knocte]

@webwarrior-ws give me an example of a type that foo could have that would fall under this _ case?

[knocte]

Other 2 testcases to add here are:

match foo with
| null -> ...
| _ -> ...

and

match foo with
| null -> ...
| notNullFoo -> ...

which both should trigger violations.

[knocte]

which both should trigger violations.

Actually, latter should only trigger violation if notNullFoo is unused (if it's used, it might prevent a race condition which would be present when using if-else block).

[webwarrior-ws]

which both should trigger violations.

Actually, latter should only trigger violation if notNullFoo is unused (if it's used, it might prevent a race condition which would be present when using if-else block).

Unused variable is detected by compiler. So perhaps it should not trigger violation?

[knocte]

@webwarrior-ws let's rebase this when you have time

[webwarrior-ws]

@webwarrior-ws let's rebase this when you have time

Rebased.