I consider this grant proposal extremely interesting, and I would love see it funded and completed.
My only suggestion would be to try and include other less known operating systems like NetBSD/FreeBSD as part of the sets of operating systems that will be tested. We know that there's a small portion of Raku users of those operating systems, but they need to be supported.
I support this as is, but think it would be nice to flesh out (or have some means for discussion of) the comment triggers. What commands do other systems offer and/or what else would we like besides the two mentioned in the proposal? Oh, and yeah, agree that some plan for *BSD and others should be mentioned. It wouldn't have to be part of this proposal necessarily, but an idea for how they could be added in the future would be good.
I think the best place to discuss comment triggers and other aspects of the planned CI infrastructure is the respective problem-solving PR: https://github.com/Raku/problem-solving/pull/219
The architecture I have in mind provides a clean abstraction of the CI backends. So adding additional backends should be trivial. (Well, as complex as the respective CIs API is.) There are CI services out there that support BSDs. (e.g. the SourceHut CI provides OpenBSD and FreeBSD workers) I'm not sure if there are easily available CI services supporting lesser known systems such as AIX or HP-UX on offer though. In every case we could always set up our own infrastructure (which brings its own can of worms with it.) I would like to limit the scope of the PR, to tackle our biggest problems first and actually finish it. We can build on that work in a next step.
This would be a great step forward and help us catch regressions or backwards compatibility problems much sooner, as in the long run it would allow for testing all published Raku modules for every commit to MoarVM/NQP/Rakudo.
Patrick is also the right person to tackle this. He has experience with all the relevant parts and demonstrated taking responsibility for the things he does.
I can contribute some iron that scales in to google cloud if that would be helpful. I personally think $4,000 is a pittance for such a project. Can you request more now or do you have to kill -9 the whole process for that?
Thanks for the offer! I'll come back to it once I'm at the point of thinking about deployment. How can I then reach you?
As this is my first grant (application), I prefer to underestimate. Once I've shown I can pull something like this off, I might aim higher the next time. But thanks for the heads up!