I'm with Mark (and Nathan and Martin) ... changing genpmda so that it
does not emit an sgi copyright assertion fixes that part of the problem.
The reference PMDA would serve a slightly different function. Most
(all?) of the PMDAs in captivity are based on the source for a
previously existing PMDA ... this stuff is sufficiently non-intuitive
that most developers will use the "copy and modify" approach to
learning. Having something that is a little more realistic than
simple/trivial that could be used as the basis for a new PMDA without
any concerns as to derivation or GPL infection would be a useful
addition if we want more people to take up PCP.
I am willing to invest the effort in making this happen ...
On Tue, 2009-03-24 at 11:01 +1100, Mark Goodwin wrote:
> Nathan Scott wrote:
> > I'd guess Kens more thinking people will find it problematic to
> > use PCP, because they don't wish (for whatever reason) to release
> > code for custom PMDAs they write and ship to their customers.
>
> Here's my 2c. genpmda was always intended to be a build tool
> rather than a run-once code generator, though I don't think it
> ever quite fullfilled that objective. In any case, IMO the tool
> itself should be GPL and copyright SGI, but the generated code
> should inherit the copyright from the config file written by
> the developer, i.e. the augmented pmns supplied with the -c option.
>
> So is generated code a derived work of the tool? In this case,
> the generated code is mostly cat <<EOF from template listings
> embedded in the tool itself. We have yacc, lex (and indeed the
> GNU compiler tool chain itself) as examples here, where clearly
> the generated code inherits neither the copyright or license
> of the tool.
>
> Cheers
> -- Mark
>
|