Fix #14844

[Clyybber]

One test will fail because the current broken behaviour allowed it to (https://play.nim-lang.org/#ix=2qAU).
Should both fail or work? A manual (var i = 0); (var i = 0) currently fails.. (for more context see bb532a6#diff-705c21e2cbb7aca092a9882547994008)

[timotheecour]

• could you add tests cases, at least the ones in Illegal symbol reuse in template with method call syntax #14844 (comment) ? (ideally turning them into doAssert tests)

• likewise with top-level vs proc scope tests mentioned there

Should both fail or work?

given that all these fail:

when defined case6:
  template bar*() =
    var r = @[(var i = @[""]; i), (var i = @[""]; i)]
    echo r
  bar()

when defined case7:
  var r = @[(var i = @[""]; i), (var i = @[""]; i)]
  echo r

when defined case8:
  proc main=
    var r = @[(var i = @[""]; i), (var i = @[""]; i)]
    echo r
  main()

it's better to be consistent; as well as to be consistent between MCS and regular call syntax, so the logical thing to do is make both fail; if absolutely needed add a -d:nimHasWorkaround14863 for a transition time.

Note that I'm fixing applyIt here to avoid the double evaluation bug: #14869

[timotheecour]

can nkEarlySemArg also be used to solve the long standing issue of const folding happening too early, eg:

• const folding happens too early in semantic phase so it's invisible to term rewriting macros and astToStr  #9301
• Don't remove information from const AST nodes in the semantic pass #6293
• Don't show const value in error message #11789
  ?

[Araq]

This is very dangerous terrain.

[Araq]

Too convoluted. Instead sem'checking should understand nkHiddenAddr, nkHiddenDeref and we need a nkHiddenArrayConstr node to make sem'checking idempotent.

[Clyybber]

That isn't related. This is about preventing typed AST ending up in untyped arguments.

[Araq]

And why is that a problem? "untyped" must go away, we need a solid typed AST so that macros can do type-based analysis. We don't need "I need untyped because the AST is then simpler", that's a symptom of the underlying issues.

[stale]

This pull request has been automatically marked as stale because it has not had recent activity. If you think it is still a valid PR, please rebase it on the latest devel; otherwise it will be closed. Thank you for your contributions.