I have many bash functions, like the example below, where the bash function unsets itself when it is done. Its has always (and still does) work. However, recently I was asked to find documentation showing this is a supported use case.

Can someone point me to a reference?

Primary environment is Debian Etch/Lenny with bash 3.x.

bash% ( set -x ; function a { echo aa ; unset a ; } ; set | grep ^a ; a ; set | grep ^a ; a )
+ set
+ grep '^a'
a ()
+ a
+ echo aa
aa
+ unset a
+ set
+ grep '^a'
+ a
bash: a: command not found

Hmm... from the bash 'help unset' command:

unset: unset [-f] [-v] [name ...]
For each NAME, remove the corresponding variable or function. Given
the `-v', unset will only act on variables. Given the `-f' flag,
unset will only act on functions. With neither flag, unset first
tries to unset a variable, and if that fails, then tries to unset a
function.

Nothing here explicitly forbids unsetting a function from within the function, and it does work. Kind of like the tapes Mr. Phelps listened to at the beginning of each episode of Mission Impossible: "Good Morning, Mr. Phelps... This tape will self-destruct in 10 seconds...."