bin/54803: /bin/sh: apparent regression with "type -p" (alias tracking?)

[david%gutteridge.ca@localhost]

>Number:         54803
>Category:       bin
>Synopsis:       /bin/sh: apparent regression with "type -p" (alias tracking?)
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    bin-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Dec 28 01:20:01 +0000 2019
>Originator:     David H. Gutteridge
>Release:        9.99.31/9.0_RC1/8.1_STABLE
>Organization:
>Environment:
>Description:
It appears a regression was introduced to /bin/sh after NetBSD 8.0 was
originally branched, such that "type -p" no longer works as it did (or,
roughly as it does in some other shells).

On NetBSD 8.0 (original release) or NetBSD 7.2:

$ type whois
whois is /usr/bin/whois
$ type -p whois
whois is a tracked alias for /usr/bin/whois

On NetBSD 9.99.31 or other recent stable or RC releases:

$ type whois
whois is /usr/bin/whois
$ type -p whois
type: usage: type name...

I'm not sure what the intent was here: if this was something that was
removed deliberately, or if it is indeed a regression?

(The reason I'm reporting this is because I was tracing why the pkgsrc
package print/lilypond was failing to build on NetBSD releases newer
than 8.0. It uses a configure script that has many examples of this:

type -p $r 2>/dev/null | tail -n 1 | awk '{print $NF}'

It's checking for the existence of various executables using this
means. Removing the "-p" of course fixes things.)

>How-To-Repeat:
As above.
>Fix:
I haven't analyzed this.