Minsoo Cheong

Some fixes are made based on review & test breaks I wondered if there should be an accept-header checking logic in httpjsontranscodingservice too but concluded that that sort of validation...

some fixes were added based on review. Plus, I missed this > If a suitable accept-header is specified, it makes sense to respect the value over content-type. should I do...

> So, we could remove [these lines](https://github.com/line/armeria/pull/4364/files#diff-bdfe29918f02ed24a4fd7e4ce1a0dc710ec18ee74f61d7d518533613d27f7b06R148-R153) completely and just pass the contentType to the frameAndServe method. I actually forgot about writing these lines 😅 removed them in commit a498c67

@minwoox Ha ha, I'm flattered 😄 But I'm going to military training in about a week, so nothing soon I'm afraid,,

@trustin That would be really nice if possible, but the provided proto files would also have to maintain their file structure in addition to their content to be fully compilable...

@ikhoon May I try the fix? since there is no standard, there's no point adding headers to com.linecorp.armeria.common.MediaType. instead, we could introduce a new class that maps custom headers to...

There does seem to be content types like `application/x-google-protobuf` or `application/octet-stream`, though maybe not as popular. ([link](https://github.com/google/protorpc/commit/eb03145a6a7c72ae6cc43867d9635a5b8d8c4545))

Adding these popular protobuf content types seem enough for this issue 🙂 But I'm still wondering whether there should be a way for users to add there own content types...

Also, I'm going to be unreachable for three weeks from now. But this fix is quite important to what we're currently doing, so if the Armeria team could do the...

> But I'm not sure that we can change the default behavior. It could be dangerous and cause serious problems in the system running with sake_case. I agree, this would...