[Records] Analyzer implementation

[itsjustkevin]

The following is a list of the individual features that need to be considered. The features are listed roughly in dependency order.

• AST enhancements (brianwilkerson, scheglov) #50087
• Element model (scheglov, NA)
• Type system updates (scheglov)
• Summary support (scheglov)
• Resolution (scheglov)
• Constant evaluation (brianwilkerson)
• Index and search (scheglov)

[itsjustkevin]

@srawlins what items are needed to call this feature work done?

[srawlins]

@scheglov @bwilkerson would you like to itemize the work? I could also go through the commit history and make a checkbox list.

[bwilkerson]

I updated it by copying pkg/analyzer/doc/process/new_language_feature.md, which might or might not be complete.

[itsjustkevin]

@bwilkerson have you found that these items cover most necessary work? Is there anything in that list I should update?

[bwilkerson]

@scheglov can confirm, but I believe that the list above is reasonably complete and that the analyzer, aside from a couple of known bugs, is complete.

[scheglov]

Yes, I think the analyzer implementation is probably complete.

However there are still discussions about parsing ambiguous code, and corresponding parser changes, and there is a bucket of co19 tests failing because of the need to disambiguate. So, there might be something new to be uncovered with tests.

[itsjustkevin]

@bwilkerson and @scheglov thank you for providing clarity around this issue!

[srawlins]

@scheglov how about now? Did the change regarding whitespace in an annotation land? And the change to make the first value in a record, $1, did that land?

[scheglov]

I believe we can mark "records" done.