Jan Schaumann <jschauma%netmeister.org@localhost> writes:
> 1) Since this is solving the same problem using the
> same input and producing the same output, the awk
> scripts there resemble those from pkgsrc/net/iana-etc/
> necessarily. Those, however, are released under the
> Open Software License ("OSL") v. 3.0. I don't know
> whether my scripts are sufficiently original to not be
> considered derived work; if not, then we'd have to
> import those scripts using the OSL.
OSL is problematic for TNF becaues it is AGPL-like and board@ decided
that was not ok for DEFAULT_ACCEPTABLE in pkgsrc, so it is obviously not
ok in base -- even though fears of having to distribute modified
services build scripts if someone does so and runs a web service are a
bit overblown :-)
If you copied enough, it's a derivative work, and if you didn't copy, it
isn't. The fact that it ends up similar is not relevant technically,
but I can see that it is a concern when assessing risk. If there's only
one way to do it sanely and you wrote it separately, i don't see an
issue, but IANAL, IANYL, TINLA as always. I looked quickly and the
programs don't look particularly similar in details.
You could write to the author and ask if they consider what you did to
be a derived work, or permission to distribute what you did under a
2-clause BSD also crediting them. I suspect there is no actual problem.
> This, then would speak in favor of leaving the tools
> in pkgsrc. The Makefile could then perform the task
> of reaching over into pkgsrc? Seems like a messy set
> up and not much of a win. :-/
I don't think the build can depend on pkgsrc. If you mean a step in a
process to update what is checked in under src, I don't see that as a
big problem.
> 2) I wanted to add the execution of the ATF tests, but
> those literally test the outcome of the files
> installed under /etc. For a full validation, one
> would need to copy the generated file into /etc, which
> seems heavy handed to me. Having a partial test that
> may generate a diff when the file is updated seemed
> reasonable to me.
It could make sense to encapsulate in ATF building our file from sources
and then comparing it to what is checked in. It seems obviously not ok
for an ATF test to modify the running system or even DESTDIR.
Attachment:
signature.asc
Description: PGP signature