run_tests.sh:2 runs rm ./tests_executable and run_tests_asan.sh:2 runs rm ./tests_executable_asan. Neither passes -f, and neither script sets set -e.
On a fresh clone the binary has never been built, so the first line of either script prints rm: cannot remove './tests_executable': No such file or directory before the build proceeds. Both binaries are listed in .gitignore, confirming they are never committed and are therefore always absent on a first checkout.
The consequence is cosmetic rather than fatal, since execution continues past the failed rm, but the noise is misleading to anyone running the suite for the first time and it leaves a nonzero status in the middle of what should be a clean run.
Suggested fix: pass -f to both rm invocations. Adding set -e to both scripts would be a reasonable companion change, so that a compile failure stops the script instead of falling through to run a binary that was just deleted, though that is a larger behavioral change and could be handled separately.
This issue body was drafted during a Gardener session (https://github.com/Stephenson-Software/gardener).
drafted by Claude on behalf of Daniel Stephenson
run_tests.sh:2runsrm ./tests_executableandrun_tests_asan.sh:2runsrm ./tests_executable_asan. Neither passes-f, and neither script setsset -e.On a fresh clone the binary has never been built, so the first line of either script prints
rm: cannot remove './tests_executable': No such file or directorybefore the build proceeds. Both binaries are listed in.gitignore, confirming they are never committed and are therefore always absent on a first checkout.The consequence is cosmetic rather than fatal, since execution continues past the failed
rm, but the noise is misleading to anyone running the suite for the first time and it leaves a nonzero status in the middle of what should be a clean run.Suggested fix: pass
-fto bothrminvocations. Addingset -eto both scripts would be a reasonable companion change, so that a compile failure stops the script instead of falling through to run a binary that was just deleted, though that is a larger behavioral change and could be handled separately.This issue body was drafted during a Gardener session (https://github.com/Stephenson-Software/gardener).
drafted by Claude on behalf of Daniel Stephenson