Closed
Description
It appears that it's not really possible to implement WorkspaceSymbolProvider if all you have available from the underlying language analysis engine is offset information (vs. line/col).
The vscode API requires that you provide line/col for every reference up front, which would require opening every file and mapping position information.
Questions:
- Could the vscode API allow passing offset information instead of line/col in this case?
- Could the vscode API allow a callback that computes line/col (after providing a TextDocument) instead of providing it up-front, since the UX doesn't need this info until after the user chooses to navigate to a reference?
- Is there any way to work around this?