getMessageAtIndex should actually return the value in the streamMessage for compatibility

[navina]

Fixing a bug from #10995

[navina]

@KKcorps : review pls!

@Jackie-Jiang should I clean up this deprecated spi- method (MessageBatch#getMessageAtIndex) before the next release?

[codecov-commenter]

Codecov Report

Merging #11355 (402a48f) into master (4c307e1) will decrease coverage by 0.01%.
The diff coverage is 0.00%.

@@             Coverage Diff              @@
##             master   #11355      +/-   ##
============================================
- Coverage     61.52%   61.52%   -0.01%     
+ Complexity     6514     6512       -2     
============================================
  Files          2233     2233              
  Lines        120083   120084       +1     
  Branches      18223    18223              
============================================
- Hits          73880    73877       -3     
+ Misses        40801    40800       -1     
- Partials       5402     5407       +5     

Flag	Coverage Δ
integration1	0.00% <0.00%> (ø)
integration2	0.00% <0.00%> (ø)
java-11	61.47% <0.00%> (-0.01%)	⬇️
java-17	61.36% <0.00%> (+<0.01%)	⬆️
java-20	61.37% <0.00%> (-0.01%)	⬇️
temurin	61.52% <0.00%> (-0.01%)	⬇️
unittests1	66.98% <ø> (-0.03%)	⬇️
unittests2	14.65% <0.00%> (+0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files Changed	Coverage Δ
...pinot/plugin/stream/pulsar/PulsarMessageBatch.java	25.92% <0.00%> (-1.00%)	⬇️

... and 10 files with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[Jackie-Jiang]

The fix looks good to me.

Not introduced in this PR, but why do we prefer reading the bytes instead of directly reading the message?

[navina]

Not introduced in this PR, but why do we prefer reading the bytes instead of directly reading the message?

The original change I introduced in #9224 was to cleanly decouple the interfaces for "fetching" a record from stream and "decoding" the payload. But that broke LinkedIn's build which was using a custom implementation of stream consumer and doesn't provide the capability to fetch raw bytes from kafka stream. So, it was fixed forward in #9544 .