Use of set -euo pipefail in Cylc generated job script

[dpmatthews]

@TomC

One issue I have is that there will be potential inconsistent behaviour if someone runs a script via Cylc versus without Cylc.

If you run a script via Cylc versus without Cylc there is no problem.
If you define some commands for Cylc to run in the job script (via script, pre-script, etc) and then you cut and paste those commands into a standalone script then, yes, you need to be careful.

@TomC

I also take issue with Cylc not being defensive and protecting itself against people doing set +euo pipefail, which is well within the rights of anyone to do. A simple change like I described earlier would add some safety to Cylc.

I don’t really see how this adds any safety.
If I understand things correctly, with Cylc 7 if you define set +e (say in env-script) then any subsequent commands that fail (non-zero return code) do not cause the job to fail unless they happen to be the last command in that section of the script definition. That makes no sense to me - why should a failure of the last command in, e.g. the pre-script be significant? I think the Cylc 8 behaviour makes more sense - if you define set +e then ignore all command failures (you are taking responsibility for error handling).