rahulgoswami commented on PR #15431:
URL: https://github.com/apache/lucene/pull/15431#issuecomment-3573820149

   Thanks for taking a look and your thoughts on this Mike. 
   
   > I think what happened is that CheckIndex is now able to read some more of 
the back-compat indexes that we previously said were incompatible. But this 
doesn't really make sense since the 10x branch does not include any additional 
backwards codecs that were removed from main. 
   
   Both main and branch_10 include backward codecs starting 8.x. The tests in 
TestBinaryBackwardsCompatibility test for binary compatibility of indexes for 
version X-2. The list of versions tested is in 
https://github.com/apache/lucene/blob/main/lucene/backward-codecs/src/test/org/apache/lucene/backward_index/versions.txt.
 It begins with 9.x for main, and 8.x for branch_10x, which explains why we 
don't hit this issue on main, but encounter this on 10x the moment we bump up 
the min version of SegmentInfos to VERSION_86 in checkHeaderNoMagic().
   
   > Some of these indexes cannot be opened, but CheckIndex is able to check 
them and reports they are clean.
   
   I think this is since CheckIndex seems to be testing for whether it can 
_actually_ read individual segments without ignoring the MIN_SUPPORTED_VERSION 
and is able to do so since the codecs are present 
(https://github.com/apache/lucene/blob/branch_10x/lucene/core/src/java/org/apache/lucene/index/CheckIndex.java#L637)


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to