[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [gopher] Updated Gopher RFC



On 9.5.2012 8:50, Mateusz Viste wrote:

I am not really a fan of such brutal connexion closing... It's what
HTTP/1.0 do (or HTTP/1.1 with connection: close header).

The nice thing about the dot is that it provides a very easy
information to the client - whether or not the menu has been
completely transferred.

Please, no top-posting - it makes this discussion very hard to follow.

Anyway, as for the lone dot, I really wasn't talking about menus at all. Gophernicus does end menus with a dot, and there's no problem with that and I'm not suggesting that we remove that.

This is the real problem:

gopher://gophernicus.org/0/archive
gopher://gophernicus.org/9/archive

The Law(tm) says that the first transfer should end with a dot and the second should not. Sounds simple enough, right? Only how in the hell is the server supposed to know whether the client expected type "0" dotted text or type "9" undotted binary? Even my Gophernicus, which by leaps and bounds the best gopher server the universe has ever seen, cannot reliably determine between the above two cases (in in above case it assumes type "0").

So, should the server break rule number #1 and never send a dot? Or should it break rule number #2 and always send the dot? I chose breaking rule #1 because that way binaries won't break...



- Kim




_______________________________________________
Gopher-Project mailing list
Gopher-Project@lists.alioth.debian.org
http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/gopher-project




Reply to: