schema for $ref keyword #211
Conversation
There was a problem hiding this comment.
OK yeah, I can support this. I did not want to put "additionalProperties": false in with "$ref" because that really is a big change, and some schemas out there (including the draft 4 hyper-schema meta-schema!) would suddenly become invalid.
I did not want to just exclude the validation properties without a lot of discussion and thought, because excluding validation keywords but not annotation or hyper-schema keywords means that we are implying that annotation and hyper-schema keywords are valid and usable in that position, and I think defining how that works is too complicated to sneak into a PR without a larger discussion. I am still very much open to having that larger discussion, preferably for the next draft :-)
While this technically makes some schemas invalid that were valid in draft 4 (since draft 4 did not validate "$ref" syntax at all), I seriously doubt anyone was relying on such schemas. So I can support this as a good tradeoff.
|
@handrews thank you. |
|
Is there an issue to discuss restrictions on $ref? |
No. You and awwright keep ignoring my requests to open one :-) (I haven't opened it because I'm not the one who wants to change it, although I am open to doing so). |
gregsdennis
added
clarification
@handrews @awwright An alternative to #194, other changes seems to be merged already