fix dot calls with resolved symbols in templates #22076
Conversation
| doAssert deb1(-12'wrap) == "-12'wrap" | ||
| doAssert deb1(-12'nonexistent) == "-12'nonexistent" | ||
| doAssert deb2(-12'nonexistent) == """(DotExpr (RStrLit "-12") (Ident "\'nonexistent"))""" | ||
| when false: # xxx bug: |
There was a problem hiding this comment.
This was not a bug. It's because this code is in a giant main template, which binds the outer 'wrap and wrap2, because that's how the language works.
| @@ -165,21 +162,16 @@ template main = | |||
| doAssert fn2() == "[[-12]]" | |||
| doAssert fn3() == "[[-12]]" | |||
|
|
|||
| when false: # xxx this fails; bug 9 from https://github.com/nim-lang/Nim/pull/17020#issuecomment-803193947 | |||
There was a problem hiding this comment.
Meanwhile this was actually the bug fixed by this PR, custom numeric literals generate dot fields which did not keep their resolved symbols in templates. It really wouldn't have taken much investigation to figure out this is related to dot fields and has nothing to do with custom numeric literals. The linked comment thinks it has something to do with 'abc not working where abc is a template param??? Very weird understanding of the language
|
Thanks for your hard work on this PR! Hint: mm: orc; opt: speed; options: -d:release |
|
@arnetheduck I think I've seen this kind of error appearing in Status' codebase. Backport to 1.6? |
|
sure, looks interesting as far as bugfixes go |
closes nim-lang#7223, closes nim-lang#11733, were fixed by nim-lang#22076
* fix dot calls with resolved symbols in templates * make old code work * fix custom number literals test * remove leftover debug marker * enable "bug 9" test too * fix renderer, add test for nim-lang#7085
add test to close nim-lang#7223, close nim-lang#11733 closes nim-lang#7223, closes nim-lang#11733, were fixed by nim-lang#22076
fixes #20073, fixes #7085
dotTransformationtransforms any resolvednkSyms as the field name back intonkIdent, but does not do this for sym choices. Generics account for this by converting tonkClosedSymChoice, but templates don't do anything similar. Now this is done in templates, but since code depended on it turning tonkIdentwhich was not resolved, we usenkOpenSymChoiceinstead.