As described in cc34919, whole idea of ever returning a parsing error in indentation engine is flawed. The code in the buffer will be incorrect even if indentation engine has no bugs. That's what people do: they write code till they decide it is complete and correct, and most of the time spent in writing it is not. If indentation engine ever stumbles upon something seemingly incorrect, it should return best approximation (typically just the previous indentation), and let the user do the rest. That's what most indentation engines do.
As part of this task we need to remove parser-error from the codebase, and once that's complete we can remove the workaround added in cc34919 as well.
As described in cc34919, whole idea of ever returning a parsing error in indentation engine is flawed. The code in the buffer will be incorrect even if indentation engine has no bugs. That's what people do: they write code till they decide it is complete and correct, and most of the time spent in writing it is not. If indentation engine ever stumbles upon something seemingly incorrect, it should return best approximation (typically just the previous indentation), and let the user do the rest. That's what most indentation engines do.
As part of this task we need to remove
parser-errorfrom the codebase, and once that's complete we can remove the workaround added in cc34919 as well.