Point "latest" meta-schemas to draft/2019-09 #297
Conversation
|
Oops. Yeah, I just forgot about this. Personally, I think that for the meta-schemas, having a "latest" that will change dramatically out from under you is a horrible idea and should never be used. It's only there for historical reasons and I'd rather remove them (but I'm guessing other people will complain? I really wish they were gone). The "latest" for the human-readable spec is OK (although we should change it to redirect if Jekyll can handle that, which it probably can't- have I mentioned how much I hate Jekyll this week?) And of course the ajv test now fails, because ajv doesn't understand @gregsdennis is it possible to run Manatee.JSON in Travis somehow? |
|
Ugh. Actually this won't work at all even with Manatee.JSON because there are relative links in the new meta-schemas so we'd have to symlink everything. @Relequestual @awwright in all seriousness can we just kill the "latest" meta-schemas? They're extremely problematic. |
|
It's .Net standard, so you should be able to run it anywhere. It'll need a console runner, though (or Powershell). |
|
@gregsdennis yeah I have no idea whether those things are available in Travis. @Relequestual do you know? |
|
maybe the site could 302 |
|
@notEthan yeah I'm looking at this. GitHub pages will only allow generating an HTML stub that will do a meta-refresh tag to redirect, which might not be acceptable for the JSON documents. Although I would still prefer it. It will be an improvement for the specs, which currently get checked in twice. Why? I have no idea, you'd think they would be symlinked... blah. |
Yeah why not. "latest" has some use for bookmarks and such, but this is probably more relevant to the documentation/specification. If you actually point to "latest" in a schema, you're (figuratively) asking for trouble. |
|
Mm. As such, I revoke my approval of this change. |
fixes #296