[gopher] Re: Item Type Suggestions

[Cameron Kaiser <spectre@xxxxxxxxxxxx>]

> We already have an URL scheme which supports mime (the gopher+ views),
> but that does not work with old clients:
> quoting rfc1738:
> 
> >    A Gopher URL takes the form:
> >
> >       gopher://<host>:<port>/<gopher-path>
> >
> >    where <gopher-path> is one of
> >
> >        <gophertype><selector>
> >        <gophertype><selector>%09<search>
> >        <gophertype><selector>%09<search>%09<gopher+_string>
> 
> 
> An url for a jpeg image made available as gophertype 9 with gopher+
> would be (e.g., if the language is set as En_Us):
> 
> gopher://server/9blablabla%09%09+image/jpeg%20En_Us
> 
> But this needs some magic to work - the client has to remove the extra
> tab, as there's no search string, because the server does not expect the
> search field to be sent when it is empty.

Yeah, about that. I'm antsy about using %09 in URLs -- not because it
won't work, and not because it's not RFC, but because I can see well-meaning
browser authors and the like throwing exceptions for a 'control character
in a URL ZOMG hack alert!!1!'. I have seen this kind of idiocy on a systemic
level. Don't think it wouldn't happen. Srsly. :)

Yes, UTF-8 requires encoding *but* it has its high-bit set, *and* it has
the weight of "everyone's using it," whereas I suspect there are very, very
few non-pathologic uses of characters < %20 in URLs in practice. We don't
even use it now for gopher; the convention has been '?' as a separator (I
don't really like that but there it is).

-- 
------------------------------------ personal: http://www.cameronkaiser.com/ --
  Cameron Kaiser * Floodgap Systems * www.floodgap.com * ckaiser@xxxxxxxxxxxx
-- Never say never again. -----------------------------------------------------