The character encoding is not specified in the extendResponse method of SseEmitter, so the character encoding actually received by the browser is ISO_8859_1.
This PR modifies the response header in the extendResponse method to encode it in UTF-8.
Should we add a constructor that accepts other character encodings?
Comment From: bclozel
I don't believe this is necessary, ISO_8859_1 is the expected encoding for servlet applications.
Comment From: YMNNs
I don't believe this is necessary, ISO_8859_1 is the expected encoding for servlet applications.
We should consider that this is an international project, and users may send information in other languages through SSE, which will cause the browser to receive garbled characters.
Another reason is that line 47 (private static final MediaType TEXT_PLAIN = new MediaType("text", "plain", StandardCharsets.UTF_8);) has specified the encoding as UTF-8, and using two encodings in the same class will destroy consistency. I believe this situation is caused by omissions in coding.
Comment From: YMNNs
@bclozel What do you think then?
Comment From: bclozel
The TEXT_PLAIN media type you're pointing is not sent with the response, but merely used to let the message converters know how they should convert the data using UTF-8.
I think server sent events should always be decoded as UTF-8 content, so there is no need to add a charset parameter to the media type. If the data is garbled, it means that you didn't serialize data properly in the first place.