Works for me!

In D55539#1273814, @kib wrote:

In particular, I believe that you should not see the problem e.g. on tmpfs, am I right?

In D55539#1272422, @kib wrote:

Are you sure that it was EINVAL? EINVAL path should not result in such condition.

I found a case where EEXISTS could do it, and hopefully fixed.

The last revision worked for me on ZFS.

• Don't use MAKE_JOBS for do-build and add an explaining comment

Is my understanding correct that it is not supposed to work with ZFS yet?

It is always a pleasure to merge a well-formed port update with a passing test suite. Kudos to Dave for preparing this change.

Makes sense to me too, but the last word is portmgr 's.

Other than last 2 comments, LGTM.

If I understand it correctly, we need "7" to be set on podman_service_rundir to allow listing its contents. Client programs usually first check for socket existence which requires the ability to list dir's contents.

This just means that I was wrong and there are no tests. Feel free to push this and I'll look into tests later.

IIRC, math/z3 supports make test. Does it pass?

What about the podman_service rc script? I guess it should get the same change?

Oh, it is a shame it still uses qt 5. Thanks for updating it, nevertheless!

After applying this patch to /usr/local/etc/rc.d/podman_service I can now run Podman Desktop without the need to do chmod on the socket. Yay!

Given that you verified that this actually works, LGTM me.

If you want an approval from me then I'd like to see the file named Makefile.cabal, not .modules.

This thing is a pain to rebase, so testing and review would be appreciated.

• Rebase

My concern is that the pointer to mp, stored somewhere in fusefs data, gets reused for other mount.

• Rebase