Right. It makes sense to splitAndTransfer underlying vector starting from startPos and not startIndex. This seems like a bug. Beyond that it looks good to me.
-Hanifi On Mon, Nov 10, 2014 at 10:01 PM, Jacques Nadeau <[email protected]> wrote: > Hey Hanifi, or anyone. > > I'm reviewing an issue with RepeatedMapVector where we need to implement > splitAndTransfer. While doing so, I ran across what appears to be a bug in > splitAndTransfer for repeated scalar values. Basically, if you look at > line 108 of RepeatedValueVectors, we transfer starting from the startIndex > position. It seems this should be based on using startPos instead of > startIndex. I understand where you corrected (rewrote/normalized the > offset values) but I don't see why you would have transferred starting at > an offset vector index rather than a value index. > > thanks, > jacques >
