Andreas Otte wrote: > Hi Dan! > > There was some time ago a discussion about utf8-ing the uri > interfaces. I think the LDAP stuff would be a customer of this. Is > this still a valid goal? I think so, yeah. But... > Aside from having an utf8 datatype for idl (bug 84816) my current > thinking goes along the following way to achive this: It sounds like the current UTF8 usage is a problem, according to jband's comments in 84186. So we're sort of stuck until someone gets time to work on that bug. The rest of the questions are someone out > > 1) Have only one escape/unescape API. Remove nsURLEscape/nsURLUnescape > (already done) and nsEscape/nsUnescape (work in progress in my tree). > > 2) Change nsStdEscape to be able to not escape chars > 127 when the > right mask (+esc_utf8) is set (done in my tree). > > 3) Have a new URIType mask entry: URI_UTF8 (done in my tree), that can > be assigned to every protocol if necessary (done with LDAP in my tree). > > 4) Store the uritype of an uri while parsing it or changing scheme for > performance reasons (Done for nsStdURL and nsSimpleURI in my tree, all > other nsIURI implementations not done). > > 5) Query uritype in nsIURI implementations when setting the escape > mask (done for nsStdURL in my tree). > > 6) Find all the others places where setting esc_utf8 makes sense ... To tell you the truth, I don't have much experience with the existing necko URI parsers, because LDAP uses a different parser altogether: nsILDAPURL / nsLDAPURL. So I'm ccing this message to the netlib group, as I think the denizens there will be able to provide better answers than I could... Dan
