Thanks Andy. Very good to hear. I think Donald is taking lead in SF last I heard. As I said not trying to dump work on you :-)
On Aug 20, 2016, at 10:11 AM, Andrew Purtell <andrew.purt...@gmail.com> wrote: With respect to my estimation of getting that SGA in about a week, I am optimistic we can turn that around. I will do what I can to drive it. There will be no problem with third party (or templates of any provenance) calling PIO code in an org.apache namespace going forward, of course. I also don't think this is particularly new ground for Apache. > On Aug 20, 2016, at 9:54 AM, Pat Ferrel <p...@occamsmachete.com> wrote: > > "a week or so” is the first I’ve heard of an order of magnitude and would > agree if that’s true. I’ll remove the suggestion but I think there is a > bigger issue here. > > Understood about what would be granted and it would not be these modified > templates but that’s that way the PIO ecosystem is meant to work. Templates > will be modified and will go back into the gallery with their “improvements” > so these templates modified to work with PIO could live side-by-side with the > donated ones (and may yet if someone else does an “improvement”). I’ve heard > nothing said here to contradict that and would be very disappointed if I did > since it is one of the bigger features of PIO IMO. It encourages this type of > ecosystem. > > BTW this is what we should hope will happen with non-apache templates. We > should be reaching out to template maintainers (I’ll do this) to get them to > use the Apache release code and add their templates to the gallery. If not, > since all the ones I’ve seen have Apache licenses, anyone is free to fork and > do these mods then put them in the gallery. > > Sorry if this makes people think in new ways, I realize the notion of > templates doesn’t have many precedents in Apache so we are on somewhat new > territory, though this has been done many times by other OSS projects. > > > On Aug 20, 2016, at 9:37 AM, Andrew Purtell <andrew.purt...@gmail.com> wrote: > > Another (minor) consideration is the donation will be by way of SGA that > specifies exact git shas. Those will be what are imported into the Apache > repos. Presumably the shas specified are not from this fork-in-progress. I > see no problem doing package munging and various other things at this time to > stage the changes for later cherry picking into the new Apache repos, though. > The SGA would grant rights to the base code and subsequent changes will be > picked over and committed by folks with ICLAs on file. > > An alternative could be to release depending on templates in the old > namespace. It's fine for projects going through incubation to not have > everything in place up front. A release in this state might pass muster. That > said, why not wait a week or so and have a grant on file, packages fixed up, > and perfect clarity? > >> On Aug 20, 2016, at 9:18 AM, Andrew Purtell <andrew.purt...@gmail.com> wrote: >> >> I don't think you should author code in the org.apache namespace that has >> not gone through proper IP clearance channels. This came up with Gearpump >> before it went into incubation. They dropped code on GitHub in the >> org.apache package namespace. I asked if that was proper. The advice >> received was it was not, because it did not originate from the foundation >> through the proper processes. We asked them to change to something like >> io.gearpump (IIRC) and they were happy to comply with the request. >> >> If the release you want to make depends on templates and those templates >> have not yet been donated and imported, I think you should wait. If you post >> a release of PIO that depends on templates in this uncertain condition I'm >> afraid I would need to vote -1 (binding) until this is resolved or we have >> clarification it's ok. >> >>> On Aug 19, 2016, at 10:09 PM, Pat Ferrel <p...@occamsmachete.com> wrote: >>> >>> No misunderstanding—so far at least :-) >>> >>> What we want to do it host the templates on Github as non-Apache projects >>> until the donation paper-work is done. This will remove a blocking >>> issue—templates that work with the new org namespace and have some (even if >>> incomplete) integration tests. We are thinking about the 7 mentioned in >>> this Jira: https://issues.apache.org/jira/browse/PIO-24 >>> >>> If you want to push them to Github we can work from that. This is >>> completely non-Apache but with Apache licenses we should be on sound legal >>> ground. PIO was designed with template forking as part of the usage model. >>> >>> If you have any problem with this, that’s fine. I’ll start making the same >>> changes you’ve done. I have no problem assuming responsibility since I >>> already maintain a couple external templates. I just want to get a release >>> of PIO done. >>> >>> >>> On Aug 18, 2016, at 3:31 PM, Chan Lee <chanlee...@gmail.com> wrote: >>> >>> Hi Pat, >>> >>> I think there may have been a misunderstanding. The work I've done is only >>> local: >>> 1) I've cloned the 10 official PredictionIO templates and changed the org >>> namespace locally. >>> 2) I've added some tests in the main repo to make sure the templates are >>> compatible with the latest release. >>> >>> As of now, I don't have write access to PredictionIO org repos, and the >>> gallery certainly shouldn't list my fork as the "official" template. So if >>> the release must happen before template donation, I would make a PR to each >>> of the template repos to change org namespace and update minimum PIO >>> version as 0.10. I will leave out the tests for now. >>> >>> But personally, I think it would be better if we wait for the legal grant >>> issue to be resolved, so that it is clearer how template code should be >>> managed. >>> >>> >>> >>> On Thu, Aug 18, 2016 at 1:13 PM, Pat Ferrel <p...@occamsmachete.com >>> <mailto:p...@occamsmachete.com>> wrote: >>> I think we are not waiting for the official template donation to release >>> PIO, can you point me to the templates you have working? I’ll make sure >>> they get added to the new gallery. We can push them to Apache once the >>> grant is done. Thanks for the help. >
