On Thursday 29 March 2007 16:51, Brett Nash wrote: > > Looks almost okay. The URL (given to a client) can start with any of the > > following, > > > > tp:// > > tps:// > > http:// > > https:// > > tphttp:// > > tphttps:// > > Cool - can we say http and https are deprecated?
Yes. Also note that as of tp04, tps will not be necessary as the tp port can support an ssl/tls filter. > > > The (.*) is there for http only I assume? (I assume it is needed for > > > http) > > > > The (.*) is the "game name". On servers which have only one game it can > > be ignored. On servers which have multiple games it should be appended > > to the username. > > That doesn't make sense? > > If the game is 'testgame' my url is: > tp://[EMAIL PROTECTED]:[EMAIL PROTECTED]:protocol/testgame I think it should be tp://nash:[EMAIL PROTECTED]:port/gameshortname. Also note that putting the password in the uri (partially on the command line) is highly discouraged. > For http servers I want the trailing data to be for the HTTP request to > get the correct location: So I don't have to only run a TP server on my > root. eg > tphttp://server/path/to/tpserver > Also how does http authentication work along with tp authentication? The tphttp and tphttps schemes are to allow using http proxy to escape tight networks. As tp is tunnelled over it, I don't think it's necessary to allow having the server not as the root. > Note that URL - two '@' and two ':' - both optional is a nightmare to > parse. I think it can't be ambiguous, but I can't prove it. One '@' and two ':' are in the standard for URIs. You don't really need more than that. > > > How does one specify a game to join if the server has support for > > > multiple games? > > > > See above. > > Once again - this needs to be better i think. And is it in @ or > trailing / format? I think it should be trailing. (I note that since tpserver-cpp only runs a single game at a time, I don't really care). > > > Should http games be specified as tp+http or similar? > > > > Do you know if tp+http is a valid name, it would be better then tphttp > > which we currently use. > > tp+http is a valid name. My understanding is[1] that it is the correct > (but not only) way of saying tp over http protocol (however you > generally register tp+http). + is a valid character for schemes. Good point, but it's not exactly correct. The git+http example is still pure http, with the path expected to be a git repo (unless it does something special, like webdav). > > > Also, is it worthwhile starting the scheme registration process for the > > > tp (and tp+http) scheme? > > > > scheme registration process? > > You can register schemes so no-one else can use it - avoids the 'dns' > issue[2]. Probably not ready for TP yet... but hopefully will one day. > Info is available in a number of places: > http://ietfreport.isoc.org/all-ids/draft-hansen-2717bis-2718bis-uri-guideli >nes-01.txt Maybe. Maybe we should register port 6923 with IANA as well. > Regards, > nash Later Lee
pgpcyRVPdBfdI.pgp
Description: PGP signature
_______________________________________________ tp-devel mailing list [email protected] http://www.thousandparsec.net/tp/mailman.php/listinfo/tp-devel
