Re: Gopher over TLS

On Mon, May 11, 2020 at 11:41 PM Sean Conner <sean-/[email protected]> wrote:

> It was thus said that the Great Sebastiaan Deckers once stated:
> > FWIW, I drafted a spec of Gopher over TLS. Published the server and
> client
> > libraries and deployed them in production.
> >
> > https://gitlab.com/commonshost/goth#gopher-over-tls-got-protocol
>
>   Not bad, I like it.  I also would *love* if the existing clients that
> attempt TLS over gopher would follow this bit:
>
>
>         If the TCP/IP socket was successful but the attempt fails without
>         receiving a ServerHello message, a GoT client may attempt to
> connect
>         without TLS, treating the connection as plaintext Gopher. This
>         failure may be cached for as long as the server's DNS records are
>         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>         valid.
>         ^^^^^
>
> There's at least one client out there (I think it's a web-to-gopher proxy)
> that attempts to ust TLS *every damn time* it makes a request and frankly,
> I'm close to just outright blocking that IP address.  Once a day (the
> current TTL for 'gopher.conman.org') isn't bad, but *every single time* is
> just rude (in my opinion).
>

Well... ehh, this is embarrassing. 🙈 Forgot to implement part of my own
spec. It's only caching the probe result for 60 seconds.
https://gitlab.com/commonshost/goh/-/blob/v4.2.1/index.js#L159
So sorry about that! Didn't mean to be a bad Gopher.

  -spc (My other pet peeve of gopher clients is the notion that every
>         selector starts with a '/' ... )
>

TIL. Did you notice that behaviour anywhere specifically? Might need to add
some tests to my Gopher repos...