Emit warnings for keywords selected to be reserved in the future

[matheusaaguiar]

Partially solves #15795 and #14770.

[cameel]

Not necessarily something to change here, but just wanted to note that our terminology is all over the place: "keyword" vs "reserved keyword" vs "reserved identifier". We should get that straight at some point, because these are not synonyms.

[matheusaaguiar]

True, I will try to align that and instead of "reserved keyword" use just "keyword" and not mix with "reserved identifiers".

[cameel]

I'd add a few more tests:

• That the future keywords trigger the warning when used as:
  • module name (import ... as X or import * as X from ...)
  • alias name (import {... as X} from ...)
  • UDVT name
  • struct/enum field name
  • Yul variable/function name
• That the future Yul keywords (i.e. leave) do trigger the warning outside of Yul.
  • It's actually up for discussion whether these should be reserved at Solidity level, but so far that has been the case with all other Yul keywords (including switch and let, which have no functionality in Solidity).
• That the future Yul reserved identifiers (e.g. blobhash) do not trigger the warning outside of Yul.

[matheusaaguiar]

I have added some tests and verified that others cover the suggestions you made. Only thing that is missing is the point about Yul variable/function name, because keywords don't seem to trigger any error when used as identifiers in inline assembly and pure Yul.

That the future keywords trigger the warning when used as:

• module name (import ... as X or import * as X from ...)

  • imports/import_name_future_keyword_warning.sol

• alias name (import {... as X} from ...)

  • imports/import_alias_name_future_keyword_warning.sol

• UDVT name

  • userDefinedValueType/udvt_name_future_keyword_warning.sol

• struct/enum field name

  • syntaxTests/enums/enum_name_future_keyword_warning.sol
  • test/libsolidity/syntaxTests/structs/illegal_names.sol
  • test/libsolidity/syntaxTests/structs/struct_name_future_keyword_warning.sol

• Yul variable/function name (?)

• Solidity Keywords (or reserved) do not trigger any error or warning in assembly or pure Yul. Not sure if that is what you were suggesting.
  see yulSyntaxTests/solidity_keywords.yul and syntaxTests/inlineAssembly/solidity_keywords.sol
  leave is already a keyword in Yul.

That the future Yul keywords (i.e. leave) do trigger the warning outside of Yul.

• test/libsolidity/syntaxTests/variableDeclaration/variable_named_leave.sol

That the future Yul reserved identifiers (e.g. blobhash) do not trigger the warning outside of Yul.

• syntaxTests/globalFunctions/blobhash_function_pre_cancun.sol (already existent)
• syntaxTests/globalFunctions/blobhash_var_shadow_warning.sol (already existent)
• syntaxTests/globalFunctions/blobhash_function_shadow_warning.sol (already existent)

[cameel]

Solidity Keywords (or reserved) do not trigger any error or warning in assembly or pure Yul. Not sure if that is what you were suggesting.

Interesting. I was not aware of that. No need to change it then. It's a bit weird, but if it's not a bug but the way things were designed then it's fine.

Still good to have that covered with tests though.

[matheusaaguiar]

They are covered. I checked and updated with some missing keywords.

test/libsolidity/syntaxTests/inlineAssembly/solidity_keywords.sol

test/libyul/yulSyntaxTests/solidity_keywords.yul

[matheusaaguiar]

@cameel , this should be ready to be merged.
( I think this was the last pending point: #16206 (comment))

[cameel]

Sure, will take a look soon. Would be good to get this one into the next release.

[cameel]

This is good for the most part. I have some change suggestions but they're minor enough (we could do away with some redundant warnings) that I think we can still merge this for the release and do them in a follow-up PR. So if you don't manage to resolve my comments today, feel free to merge and do just that.

Just make sure to apply the changelog suggestions.