On Thu, Jun 11, 2009 at 11:15:14AM +0200, Peter Rosin wrote: > Den 2009-06-11 11:48 skrev Daniel P. Berrange: >> On Thu, Jun 11, 2009 at 10:38:45AM +0200, Peter Rosin wrote: >>> Den 2009-06-04 13:34 skrev Pierre Ossman: >>>> So to summarise, the only part that has some idea of what encoding it >>>> wants is the Unix client. Everything else is basically random and >>>> dependent on the system settings. ASCII should be fairly safe to use, >>>> but I'm not 100% sure of even that. >>>> >>>> Given that most of the systems (except for the Unix server) have been >>>> using 8859-1 (or cp1252, which is close) in practice, I'd say we define >>>> the desktop name to be 8859-1 and nothing else. >>> I say we define it as ASCII and have existing code continue working. If >>> you want anything else, use the DesktopName encoding (which I think we >>> should define as UTF-8). >> >> Or how about we define a psuedo-encoding/extension for the purpose of >> negiotiating a character set. Once the client/sever agree on a charset >> it would then apply to all other extensions/messages containing strings. >> IMHO it makes more sense to have the same charset for all messages, and >> switch the encoding of the whole connection in one go, than to define >> different charsets for different messages. > > That will have to be a brand new security type as pseudo-encodings are > after the ServerInit message, which makes it cumbersome. Or? Plus, > anything other that UTF-8 seems pretty unreasonable. Or did you mean > that all servers (or clients) *needs* to contain a bunch of conversion > functions (or add an extra conversion dependency) for various esoteric > character sets?
I was thinking any charset negotiation, but you're right that it is really pretty pointless. Just need to get client/server to agree on UTF-8 and be done with it. > Having a security type named UTF-8, which simply allowed the client > to select the "real" security type right after it would perhaps not be > so difficult. I've never much liked the idea of stealing security types to negotiate non-security related features myself. It is pretty trivial to implement the DesktopName encoding. So if we did it with a pseudo-encoding, the serverinit message would still be ASCII, but would recommend that the server send a DesktopName message with the UTF-8 encoded name version. Daniel -- |: http://berrange.com/ -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://freshmeat.net/~danielpb/ -o- http://gtk-vnc.sourceforge.net :| ------------------------------------------------------------------------------ Crystal Reports - New Free Runtime and 30 Day Trial Check out the new simplified licensing option that enables unlimited royalty-free distribution of the report engine for externally facing server and web deployment. http://p.sf.net/sfu/businessobjects _______________________________________________ tigervnc-rfbproto mailing list tigervnc-rfbproto@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/tigervnc-rfbproto