Question for PriorityQueue Reconstruction in SloppyPhraseMatcher's advanceRpts method.

https://github.com/apache/lucene/blob/33a4c1d8ef999902dacedde9c7f04a3c7e2e78c9/lucene/core/src/java/org/apache/lucene/search/SloppyPhraseMatcher.java#L311

Hi, for this line of code, I got confused on this PriorityQueue Reconstruction step, shouldn't it check that the popped pp2 is in the same rptGroup as the original pp? Why it only checks rptGroup>=0? The popped pp2 can be from another rptGroup so that the real resolved pp never got popped out.

Or there is some theory that guarantees if we pop n elements and push them back, then the heap can be correctly re-constructed? (I came up with some simple counter-examples so I guess there is no such a theory? Please correct me if there is one)

Thanks a lot!

Start at lucene/core/src/java/org/apache/lucene/search/SloppyPhraseMatcher.java around line 311 and trace advanceRpts, the PriorityQueue reconstruction, and rptGroup handling. Compare the heap behavior with the issue's counterexamples and relevant Lucene search tests. Done means establishing whether the reconstruction is correct and, if not, documenting the failure and the required correction.

Written by the indexing model from the issue text.