Current situation:

When  we import a depencency from fedora srpm which is using rpmautospec, we get %autorelease as Release value but also we have the definition of autorelease dynamically added into the spec. i.e.:

{{}}

## START: Set by rpmautospec

{{## (rpmautospec version 0.3.5) }}

{{## RPMAUTOSPEC: autorelease, autochangelog }}

%define autorelease(e:s:pb:n) %{?-p:0.}%{lua:

  release_number = 1;

  base_release_number = tonumber(rpm.expand("%{?-b*}%{!?-b:1}"));

  print(release_number + base_release_number - 1);

}%{?-e:.%{-e*%{?-s:.%{-s*}}%{!?-n:%{?dist}}}}

## END: Set by rpmautospec

{{}}

[...]

Release: %autorelease

So we don't need rpmautospec and we are actually not running it at all but reusing the values that rpmautospec defined in fedora.

Some docs in https://docs.pagure.org/fedora-infra.rpmautospec/ 

The problem:

If we use the spec we import in a mock o koji instance with support for rpmautospec, the plugin detects that rpmautospec is required by the mere presence of "%autorelease" macro and will try to install it and run it again, which is not the behavior we desire.

Proposed solution:

To avoid this situation, I think we should avoid using %autorelease macro when importing deps.