mlevkov opened a new pull request, #4160:
URL: https://github.com/apache/iggy/pull/4160

   Closes #4159.
   
   The integration harness resolves the binaries it launches with 
`Command::cargo_bin`, which finds whatever is already in `target/` and never 
builds it. CI does not run into that because it builds first, in 
`_test_bdd.yml`, `coverage-baseline.yml` and several per-language actions. The 
command written in this file for running a single integration test does not, so 
a binary left over from an earlier checkout gets used without a word.
   
   The cost is not the rebuild, it is the time spent reading the failure. In my 
case a five day old `iggy-server` failed 22 tests across two suites, the plain 
happy path among them, with a panic naming a config file and a derive macro 
that were both fine, on a commit whose CI was green. Everything about that 
reads as a broken change rather than a stale binary.
   
   Two small changes. The Testing section builds before it tests, so the 
snippet people copy is correct on its own. Pitfalls carries the general rule, 
since that section already collects traps of this shape.
   
   Documentation only, no behaviour change. #4159 also describes an optional 
follow-up: `TestBinaryError::StartupTimeout` is the one place that already 
knows the server did not come up, so including the binary's modification time 
there would make the message actionable without guessing. I did not do that 
here, since it changes shared harness code and seemed worth a maintainer's 
opinion first.
   


-- 
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]

Reply via email to