Re: fixing coda in -current

[maya%netbsd.org@localhost]

On Wed, Nov 21, 2018 at 08:04:20AM +1030, Brett Lymn wrote:
> On Tue, Nov 20, 2018 at 07:39:26PM +0000, maya%netbsd.org@localhost wrote:
> > On Tue, Nov 20, 2018 at 08:06:37PM +0100, Hauke Fath wrote:
> > > ISTR that somebody on the CODA mailing-list suggested a re-implementation
> > > as userland file-system, but I don't think much has happened on that front.
> > 
> > note that you can cd /usr/tests/fs; atf-run as user, because
> > they're mostly not running on the kernel :-)
> 
> You need the userland installed from pkgsrc and a coda server configured
> before it will work so setting up atf is challenging.
> 
> I do have some private code that provides a mount_coda so you can have a
> fstab entry but it still relies on pkgsrc being there to work.

The pkgsrc aspect doesn't seem insurmountable: we could check if
the needed tools are installed, and skip if they aren't. Then, create a
meta-pkg netbsd-tests for installing all the tools possibly ever needed
by the testsuite from pkgsrc.

Similarly, we have meta-pkg/netbsd-www for editing the website and it's
very comfortable to use.

Not sure how to overcome it being a network filesystem needing a server.
I wonder if they have their own tests to be inspired by. Anita+qemu is a
thing, but that makes it longer and heavier than typical tests.

Is our code compatible with e.g. linux clients?