First summary: passwd package should be essential? It probably shouldn't.
So far, we only got two advices but, imho, enough motivated to make me
change my initial feeling.
It seems that nothing has yet motivated that passwd should indeed be
Steve bringed the very interesting rationale: "I think we really
should not be using it *except* for packages that we require to be
functional when in an unconfigured state. The passwd package
certainly doesn't qualify in this". He's right: passwd is perfectly
functional in unconfigured state.
He also counters the argument of paswd utilities being needed in
config scripts by explaining that packages requiring
useradd/userdel/etc in *config* scripts are probably wrong.
Lars added mostly the following: "Is there a problem with packages
that need stuff from passwd simply depending on passwd".
He also seems right. There doesn't seem to be any problem to this as
long as the requirement is not in config scripts. Moreover, most
package who would depend on some passwd stuff probably would because
they need to add/remove users or groups. However, a recent survey has
proven that indeed nearly all packages doing this actually (Pre-)Depend on
adduser and use the high-level utilities in adduser rather than
low-level utilities from passwd.
For the above reason, some of these package may be indeed broken if
they require either adduser or useradd in their config script. But
that's these packages problem not passwd problem.
In summary, it will need a lot more advices following Kurt Roeckx
suggestion in #345651 to change my mind back and make passwd
Kurt, would you mind commenting?