See PR #3167 for a lot of context.
In OAS 3.0.3:
It's clear that format: byte was intended to carry over from OAS 2.0, but that base64 is more intuitive and got used everywhere outside of the table of formats.
@OAI/tsc I feel like the most sensible thing to do here would be to define both byte and base64 as having the same effect, and note that byte is compatible with OAS 3.0, while base64 is analogous to using contentEncoding: base64 in OAS 3.1.
I'm guessing that adding a format would not be allowed by SemVer, but we could make the argument that the current situation is so confusing that defining both is just a clarification.
See PR #3167 for a lot of context.
In OAS 3.0.3:
format: bytefor base64-encoded binary data, matching the same section in OAS 2.0format: base64alongside an example offormat: binary; there is no sign offormat: bytemultipartContent mentionsformat: base64alongsideformat: binarywhen talking about defaultContent-Typevalues, but does not mentionformat: byteIt's clear that
format: bytewas intended to carry over from OAS 2.0, but thatbase64is more intuitive and got used everywhere outside of the table of formats.@OAI/tsc I feel like the most sensible thing to do here would be to define both
byteandbase64as having the same effect, and note thatbyteis compatible with OAS 3.0, whilebase64is analogous to usingcontentEncoding: base64in OAS 3.1.I'm guessing that adding a format would not be allowed by SemVer, but we could make the argument that the current situation is so confusing that defining both is just a clarification.