Ability to specify type to 'zfs destroy`

[mailinglists35]

System information

Type	Version/Name
Distribution Name	all
Distribution Version	all
Linux Kernel	all
Architecture	all
ZFS Version	all
SPL Version	all

Describe the problem you're observing

There are no safeguards when destroying. Whatever is the argument of zfs destroy, it will get destroyed. Issuers of the destroy command are humans and humans can make mistakes: for example in shell scripts that generate destroy commands, typos etc.

A minimal protection mechanism could reject a destroy action if the type of the dataset to be destroyed does not match the dataset type specified in command line, similar to how zfs list -t works, so for example zfs destroy -t filesystem pool/volumename or zfs destroy -t snapshot pool/filesystem would fail.

Describe how to reproduce the problem

Include any warning/errors/backtraces from the system logs

[paranoidi]

This was my pretty much first thing I noticed about ZFS. I'm currently checking it out since I fear the day I accidentally delete my files and snapshots provide excellent guard against that. However the destroy command looks a bit like ACME product ...

Edit: After some thinking, one possible solution that does not require backward incompatible CLI changes and does not rely user remembering to specify -t type would be to add "destroy guard" option flag on the pool.
Edit 2: Disregard first edit

[mailinglists35]

changing behaviour of defaults and/or existing options IS a backwards compatibility breaking, adding a new option is not.

adding new options is how evolution works while keeping backwards compatibility

you don't need a feature flag for pool, because the change i'm requesting is a behaviour enhancement that has nothing to do with how the on-disk format is layed out.

in other words, this can be implemented in zfs command without needing to alter the pool attributes. altering the pool is unneeded for this particular case and will only complicate the administration of zfs.

a script that wants to use this feature would have to simply invoke zfs --version and determine if the -t option is available or not by knowing in what release this option was made available. a human would have to do the same, or look up the option in the manual page of the target installed system.