Brian, Yes, the resulting JS string will see two characters representing the UTF-16 surrogate pairs in UCS-2. Nothing was changed in Node for this, it was all V8. I only mentioned it in the ChangeLog because I know that many Node users were affected by it.
On Fri, Apr 6, 2012 at 12:18, Brian <theturtl...@gmail.com> wrote: > I saw the thumbs-up on my iPhone. :) > > As the author of WebSocket-Node, this has been a bit frustrating since no > Node-based WebSocket implementation can legitimately pass the UTF-8 > validation tests from the ubiquitous AutoBahn Test Suite - > http://autobahn.ws/testsuite > > How was this accomplished exactly? Surrogate Pairs? I was under the > impression that V8 itself was unable to properly represent non-BMP > characters because it uses UCS-2 internally. There was an extremely recent > commit to support non-BMP characters by using UTF-16 surrogate pairs, though > the JS string methods would see those surrogate pairs as two characters > rather than one. See http://code.google.com/p/v8/issues/detail?id=761 > > Brian > > > > On Thu, Apr 5, 2012 at 10:27 PM, Alan Gutierrez <a...@prettyrobots.com> > wrote: >> >> >> >> On Thursday, April 5, 2012 at 5:41 PM, Bert Belder wrote: >> >> > On Apr 4, 8:08 am, Matt Ranney <m...@ranney.com (http://ranney.com)> >> > wrote: >> > > 👍 >> > > >> > > I wish I could paste a one-character emoji response here to >> > > efficiently >> > > express how happy I am about this, but Chrome/Gmail doesn't support >> > > non-BMP >> > > characters either. >> > >> > >> > >> > Well, the code point of the character in your message seems to be >> > correct (U+1F44D THUMBS UP SIGN). So it seems more likely that you >> > have no font for it. >> >> >> I have the font for it. I see the thumbs up. I feel the joy, the triumph. >> Alan > >
