-
Notifications
You must be signed in to change notification settings - Fork 29
Description
Looking at https://github.com/apache/datafusion-sqlparser-rs, a very requested feature from community that is almost fully implemented (think there are still som edge cases and areas lacking coverage - but getting there.
Now, I understand that a Polyglot AST that is transformed, it doesnt really make that much sense to maintain spans as they are going to move and shift around, (and even more so, writing to an other dialect would clearly make it "off").
I was just using the playground and i.e for stuff like parsing errors its super hard to find the error if the statement is non-trivial. I.e reporting back a red squiggly line or marking a red dot next to line number.
An initial iteration could at least target just the tokenizer side of things so we could report clear syntax errors.
A more advanced one could somehow "hold on" to the "original" token, so we could use that to i.e report a type error or semantic analysis/ validation error.