Den 2009-06-29 11:16 skrev Pierre Ossman: > On Fri, 26 Jun 2009 21:15:40 +0200 > Peter Rosin <p...@lysator.liu.se> wrote: > >> Den 2009-06-25 16:26 skrev Pierre Ossman: >>> Incompatible is a term that will have to be used loosly here. There is >>> no compatibility here today as everyone is doing different things, so I >>> don't see this move making things any worse. >> I never answered this, so here I go. I'd be happier if it was >> formulated more like a recommendation (a strong one if you >> like) to use UTF-8 whereever the encoding was unclear. As the >> patch is written it seems like an error to not use UTF-8. >> >> In particular this snippet is bad: "Clients and servers must >> send UTF-8 strings". "Must" is just way too strong. >> > > I originally wrote "should" there, but that gives the impression that > things work fairly fine even without following this behaviour, which it > doesn't. > > It's a rather odd situation we're in. Not having a defined encoding > makes the strings nearly useless, which warrants a "must". But OTOH, > that's how it is today and therefore a "should" would normally be > appropriate. > > IMO it's better to have a strong "must" as all strings really need to > have a defined encoding, but keep the historical note to warn against > the horrible situation we're in with the current implementations.
And I think "should" is also too strong. It should say that the encoding *is* unspecified and that everybody "must" treat strings with care because of this, and then go on to "strongly recommend" everybody to use UTF-8 (as that's the only sane choice) or limit themselves to ASCII if they are willing to trade usability for portability. IMHO of course. If you want to use the words "must" or "should", you need a UTF-8 extension. >> I also think the second hunk of the patch should be left out >> so that the string encoding is left unspecified and thus falls >> back to the new "String Encodings" section (which of course >> recommends UTF-8). > > It was for redundancy as I suspect many browse through the initial > parts of the spec very quickly. Just put in a reference to the "String Encodings" section instead. Cheers, Peter ------------------------------------------------------------------------------ _______________________________________________ tigervnc-rfbproto mailing list tigervnc-rfbproto@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/tigervnc-rfbproto
