Looking at https://gitbox.apache.org/repos/, I see several projects,
perhaps most projects, that have a special repository just to contain
testing logic. Often this is named <project>-testing and often is has
other names (like integration and others).
So there is a precedence for a nuttx-testing repository to hold
testing data... Things like yml files and the like. I would much
prefer to a have a separate repository to hold test files than to
contaminate the incubator-nuttx and incubator-nuttx-apps repositories
with test data.
Another thing that I note by googling ".gitmodules testing
site:github.com/apache" is that many of these testing directories also
contain submodules to coordinate testing. I have gotten to rather
like the idea of three repositories: incubator-nuttx,
incubator-nuttx-apps, and incubator-nuttx-testing that holds all of
the test logic with we really don't want to contaminate the user
facing repositories with. Best to put testing logic in one place and
keep the end-user repositories as clean as possible to support the
user needs.
I know, I am violating my own principles. We should not really be
discussing testing repository organization UNTIL we have established,
agreed-upon workflow requirements. We don't have any such requirements
so just think of this as idle talk. For me it clarifies my position on
this. I am not in favor of any mechanization that keeps the end-user
repositories and clean as possible.
Greg
