[tcgc] spread behavior change

[tadelesh]

use original model for body parameter in SdkServiceOperation if it is from a simple spread.

simple spread means spread model without any @header, @query or @path property and the spread model is the only source of http body.

examples,

op test(...Model): void;

op test(@header h: string, @query q: string, ...Model): void;

these two operation will have Model as body parameter.

op test(...A): void;
model A {
  @query
  q: string;
  prop: string;
}

this operation will have an anonymous model with only prop property and generated name TestRequest as body parameter.

resolve: #1132

cc: @ArcturusZhang @joheredi

MIGRATION GUIDE

In previous version, we could easily judge if a body type is spread by:

function isSpreadBody(bodyParam: SdkBodyParameter | undefined): boolean {
    return bodyParam?.type.kind === "model" && (bodyParam?.type.usage & UsageFlags.Spread) > 0
}

Now, since the spread model could also be used by other operations, you need to change to the following judgement:

function isSpreadBody(bodyParam: SdkBodyParameter | undefined): boolean {
    return bodyParam?.type.kind === "model" && bodyParam.type !== bodyParam.correspondingMethodParams[0]?.type;
}

[azure-sdk]

All changed packages have been documented.

• ✅ @azure-tools/typespec-client-generator-core

Show changes

@azure-tools/typespec-client-generator-core - feature ✏️

use original model for spread if it is from a simple spread

[azure-sdk]

You can try these changes here

🛝 Playground	🌐 Website	📚 Next docs

[tadelesh]

@ArcturusZhang @joheredi please review and check if it could satisfy your language's requirement, thanks.

[ArcturusZhang]

Do we have a dev version so that we could see if this change could meet our requirement?
The code here LGTM

[tadelesh]

Do we have a dev version so that we could see if this change could meet our requirement? The code here LGTM

before this pr merge, you could use the ci build artifact. after merge, you could use the dev version.

[tadelesh]

@ArcturusZhang have you tried this in .net emitter? if all behavior is good, could you approve this pr?