Watching a token for renaming

Hi Matt,

No we don't have anything that tracks that.  I would probably suggest you watch the document text changed event to see if typing occurred, and see if you are on an identifier.  If so, you could possibly look at the undo history.  You should have full access to the text change operations in each text change in the undo history.  So you could theoretically walk those, combined with snapshot examination, and determine if typing was in an existing token or not.

Ok, so the methodology I used was to combine the derivation of the type of selection change event from the other thread (http://www.actiprosoftware.com/community/thread/23346/determining-cause-of-ieditorviewselectionchan) and inspecting the token under the cursor if it wasn't a typing change, and storing the text range and token Id.

Then, if it was a typing change and a watched token existed, I brought the text range forward to the current snapshot with ExpandLastEdge, and checked:

1) whether the typing occurred within the text range of the watched token

2) whether the watched token represented the same type (i.e. I hadn't written keywords etc when changing an identifier)

If both of those conditions are true, I put up a smart tag session activated with Ctrl+Period to enable refactoring.

If it was a selection only change, then I simply checked whether the caret was still within the text range of the watch token's text range (again bought forward to the current snapshot with ExpandLastEdge). If it wasn't in that range, I cleared the watched token.

That works pretty well for me, so I thought I'd let you know in case that sort of functionality is something you'd consider worthwhile when implementing smart tags yourself...

Thanks Matt, I'm glad it's working well!