Something we've been handling very cavalier so far is: when does a schema file get uploaded to the xsd repository?
I think there's no need to upload schemas during WD; early adopters can be expected to grab them from github (or an auxiliaryurl in the document) and somehow shoehorn them into whereever they need the schema.
Once you're doing RFC... Well, on the one hand I think it would be great if the community could already rely on the namespace URI to resolve to the new version. On the other hand, that thing then is still under review and might change incompatibly, which is a strong argument against handing it out from an "official" IVOA URL. Again against that: Since the old and then new XSD share a namespace URI, they are supposed to be compatible in the sense of everything being old-valid also needs to be new-valid. By handing out the new version early on, we might spot regressions before the schema becomes official...
Oh, and then there's the problem of the version attribute on the schema root... I'm currently recommending in ivoatexDoc that people add tags like -wd3 or -pr1 to its value, but that XSDs coming with RECs (and ENs) must not have such tags. I believe that's by and large a good idea, and I think DocStd is where we should require it.
Something we've been handling very cavalier so far is: when does a schema file get uploaded to the xsd repository?
I think there's no need to upload schemas during WD; early adopters can be expected to grab them from github (or an auxiliaryurl in the document) and somehow shoehorn them into whereever they need the schema.
Once you're doing RFC... Well, on the one hand I think it would be great if the community could already rely on the namespace URI to resolve to the new version. On the other hand, that thing then is still under review and might change incompatibly, which is a strong argument against handing it out from an "official" IVOA URL. Again against that: Since the old and then new XSD share a namespace URI, they are supposed to be compatible in the sense of everything being old-valid also needs to be new-valid. By handing out the new version early on, we might spot regressions before the schema becomes official...
Oh, and then there's the problem of the version attribute on the schema root... I'm currently recommending in ivoatexDoc that people add tags like -wd3 or -pr1 to its value, but that XSDs coming with RECs (and ENs) must not have such tags. I believe that's by and large a good idea, and I think DocStd is where we should require it.