Fix count when deleting compression chunk size entry

[erimatnor]

In the function ts_compression_chunk_size_delete() a counter is used to indicate the number of catalog entries deleted. But this counter is never incremented so the returned count is not accurate.

Fix this issue by incrementing the counter when entries are deleted. Also, increment the command counter (to make changes visible) only at the end if the count is non-zero.

Disable-check: force-changelog-file

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.24%. Comparing base (59f50f2) to head (e69c974).
Report is 359 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #7297       +/-   ##
===========================================
+ Coverage   80.06%   92.24%   +12.18%     
===========================================
  Files         190      205       +15     
  Lines       37181    38565     +1384     
  Branches     9450     9999      +549     
===========================================
+ Hits        29770    35576     +5806     
+ Misses       2997     2989        -8     
+ Partials     4414        0     -4414     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[fabriziomello]

Is possible to add some tests either regression or unit for this change?

[erimatnor]

Is possible to add some tests either regression or unit for this change?

Note sure what kind of tests you'd like to see. This is just fixing an issue with a count that wasn't correct. Nothing is currently looking at the returned count. The code functionally is the same as before.

[akuzm]

Note sure what kind of tests you'd like to see. This is just fixing an issue with a count that wasn't correct. Nothing is currently looking at the returned count. The code functionally is the same as before.

I can think of two things:

1. You added CommandCounterIncrement, this means that something is now visible that was not visible before, probably there can be an isolation test to detect this?
2. If nothing is looking at the returned count, why are we returning it? I'd remove it as dead code then.

[fabriziomello]

Is possible to add some tests either regression or unit for this change?

Note sure what kind of tests you'd like to see. This is just fixing an issue with a count that wasn't correct. Nothing is currently looking at the returned count. The code functionally is the same as before.

I don't know if it is possible or no because I'm not too familiar of this part of the code. So every time that a pick to review something that I'm not familiar I start looking to the tests to try to understand the change.