Bump cadl ranch 0.37.1 #2815
Bump cadl ranch 0.37.1 #2815
Conversation
| // for `temperature: HttpPart<{@body body: float64, @header contentType: "text/plain"}>`, the real type is float64 | ||
| sourceType = body.type; | ||
| } | ||
| } |
There was a problem hiding this comment.
For temperature: HttpPart<{@body body: float64, @header contentType: "text/plain"}>, TCGC provide model of {@body body: float64, @header contentType: "text/plain"} while python emitter actually only need float64 so we need to parse it.
There was a problem hiding this comment.
is it bad that we have contentType? I'm thinking we still just pass it through. But I'm good either way
There was a problem hiding this comment.
HttpPart permit us to define more specific info for this part. In this case @header contentType: "text/plain" means SDK shall serialize the float64 as plain string instead of json string. Although both actually send same content, JS owner thought we shall add the specific info in the cadl-ranch case then I just added it.
There was a problem hiding this comment.
I raise an issue Azure/typespec-azure#1488. If all languages agree Azure/typespec-azure#1488 (comment), TCGC could do the parse logic so that emitters don't need the duplicated work any more.
| return ( | ||
| self.type.is_form_data | ||
| or bool(self.entries) | ||
| or ("multipart/form-data" in self.content_types and self.code_model.options["from_typespec"]) |
There was a problem hiding this comment.
why do we need to check if it's "from_typespec". We should also remove this option and deal with the differences earlier on in the pipeline
There was a problem hiding this comment.
Without from_typespec, there will be diff for legacy code under autorest.python folder. Since multipart is mainly for typespec design, I think we had better keep legacy same as before.
| // for `temperature: HttpPart<{@body body: float64, @header contentType: "text/plain"}>`, the real type is float64 | ||
| sourceType = body.type; | ||
| } | ||
| } |
There was a problem hiding this comment.
is it bad that we have contentType? I'm thinking we still just pass it through. But I'm good either way
…bump-cadl-ranch-2024
No description provided.