lets-order-some-fries commented on PR #68096:
URL: https://github.com/apache/doris/pull/68096#issuecomment-5873650171

   Thanks @morrySnow for running buildall — green across the board.
   
   Both of the automated review's requests are addressed in `5fd32d86` 
(test-only):
   
   1. **UPDATE / DELETE keep the ordinal's digits.** `LogicalPlanBuilderTest` 
parses `ORDER BY 4294967297`
      and `ORDER BY 18446744073709551617` in both statements and asserts the 
`UnboundSlot` is named
      exactly after the literal.
   2. **LARGEINT width, and both ORDER BY paths.** The GROUP BY test now also 
uses
      `18446744073709551617` (2^64 + 1), which narrows to 1 through 
`getLongValue()` as well. A new test
      covers ordinary and set-operation ORDER BY on the analyzed plan: a wide 
ordinal must never become a
      sort on a column, with `ORDER BY 1` as the control that shows the check 
can see the difference.
   
   @yx-keith — (2) is also the plan-level sort-key check you asked for on 22 
Sep. I held it back then
   because I couldn't run it; I can now.
   
   **This time it ran locally, on a full `fe-core` build:**
   
   - `LogicalPlanBuilderTest` + `BindExpressionTest`: 30 tests, 0 failures; 
`checkstyle:check`: 0 violations
   - with the fix reverted to `getIntValue()`, all three tests fail —
     `DELETE FROM t ORDER BY 4294967297 LIMIT 10 ==> expected: <4294967297> but 
was: <1>`
   - with `getLongValue()` in both places, all three still fail, on the 
LARGEINT value only —
     `expected: <18446744073709551617> but was: <1>` — which is the gap the 
review pointed out
   
   Could someone re-run `/review` so the blocking review reflects the new head, 
and `run buildall` for the
   added tests?
   
   @CalvinKirs — thank you, that was the missing piece. The `automation` 
package ships thrift 0.24.0,
   which took the FE reactor from 25 modules to 51 here. The last blocker on 
Apple Silicon turned out to
   be `protoc-gen-grpc-java`: the `osx-aarch_64` binary Maven Central publishes 
for it is actually an
   x86_64 executable, so without Rosetta `fe-grpc` fails with *"program not 
found or is not
   executable"*. A native build (Homebrew's `protoc-gen-grpc-java`) passed in 
through
   `-Dgrpc.java.artifact=...` gets the whole FE building — in case it saves the 
next arm64 contributor an
   evening.
   


-- 
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