>> Please be a bit more specific.
>>
>> As I understand it, the problem I am observing is caused by pkgin
>> not adhering to the declared "remote" pkg dependencies.
>>
>> The web page above about the location of the local pkg database
>> and its move to /usr/pkg/pkgdb/ only affects the *local* storage
>> of package install list and pkg dependencies.
>
> I picked up a hint, perhaps incorrectly, that you had pkgdb confusion.
>
> I am guessing that pkgin does not directly access PKGDB_DIR. But if it
> does, that's another source of potential trouble.
Here's some counter-evidence:
# ktrace -i pkgin up
processing remote summary (https://cdn.NetBSD.org/pub/pkgsrc/packages/NetBSD/amd64/10.0/All)...
pkg_summary.bz2 100% 4225KB 384.1KB/s 00:11
# kdump | grep /pkgdb
"/usr/pkg/pkgdb\n"
"/usr/pkg/pkgdb\n"
22219 22219 pkgin NAMI "/usr/pkg/pkgdb"
22219 22219 pkgin NAMI "/usr/pkg/pkgdb"
#
But as you can see, these all refer to the "new" pkgdb location.
Also, I have no /var/db/pkgdb/ directory.
Also, the "in-tree" pkg_add only refers to the new location:
# strings -a /usr/sbin/pkg_add | grep /pkgdb
/usr/pkg/pkgdb
/usr/pkg/pkgdb
#
The in-tree pkg_install is from NetBSD/amd64 10.0, and is:
# type pkg_add
pkg_add is /usr/sbin/pkg_add
# pkg_add -V
20210410
#
and evidently, none of the "native" pkg_install nor the pkgsrc
pkg_install (or pkgin, for that matter) depend on
pkg_install.conf (in either location) to use the same and "new"
pkgdb location.
So I think we can disregard the possibility of "differing pkgdb
locations" as being the explanation for the issue I was seeing.
I'll proceed to test the patch Jonathan Perkin developed to see
if that is a proper fix for the issue (on the face of it, it
looks promising).
Best regards,
- Håvard