Re: reset a trap upon use

[Koichi Murase]

2023年11月25日(土) 1:16 bill-auger <bill-auger@peers.community>:
> [...]
> just wanted to know if this is the expected behavior (or did my predecessor 
> write a bug
> by placing the trap reset inside of the cleanup() function)

It is the expected behavior.  The traps for DEBUG, RETURN, and ERR
pseudo-signals are maintained for each function call separately.
Unless `set -T' (for DEBUG and RETURN) or `set -E' (for ERR) is
specified or the trace attribute (`declare -t') is specified to the
function (for DEBUG and RETURN), those traps for the pseudo-signals
will not be inherited by the child calls of the function.  On the
other hand, non-empty traps will be propagated to the callers.  To
make a function inherit the RETURN trap of the caller and propagate
the unset RETURN to the caller, you need to mark the function with the
trace attribute.  More specifically, you need to mark your `cleanup'
as

declare -ft cleanup

--
Koichi