This should be pretty straightforward. Any volunteers?

I think I would implement the underlying functionality only in WTF::TextCodecUTF8 rather than add a virtual method to WTF::TextCodec which would only be implemented for UTF8 anyway.

Not iOS (which does not use Blink)

Tagging with GFB to hopefully catch some attention. 

This will require an "Intent to Implement and Ship"; happy to help with that if an external contributor picks this up.

Interested in taking it, but can't yet guarantee an timeline. Will keep this thread updated though.

Following ricea@'s advice, my thoughts were to make a method similar to WTF::TextCodecUTF8::Encode that takes in a "capacity" in addition to the "length". Then we  can make something like WTF::TextCodecUTF8::EncodeCommon that takes in the capacity, and calls U8_APPEND with it, instead of U8_APPEND_UNSAFE. However, this requires a TextCodecUTF8::Encode method with a new signature, and thus a change to WTF::TextCodec to support this IIUC.

Thoughts on getting around this to avoid touching all of the TextCodec subclasses?

I think it would be acceptable to have virtual WTF::TextCodec::EncodeInto() { NOTREACHED(); }, overridden only in TextCodecUTF8. And then WTF::TextEncoding::EncodeInto() with a sufficient warning that only UTF8 supports it.