[standards-jig] JNG Ramblings.
m at tthias.net
Fri Aug 9 18:19:38 UTC 2002
Thomas Muldowney wrote:
>I think the need/want for in band binary data is the key to this
>argument. If you don't want in band binary data, obviously you don't
>want any binary information in the protocol. Is that potentially why a
>binary frameing protocol doesn't sit well with you? Or, would it still
>sit bad if the protocol had in band binary?
If the protocol had in band binary I'd prefere to use an existing
protocol over inventing a new one. - And BEEP can be used for that ...
and gives also other additional possibilities like virtual SSL hosting.
But I think nearly-inband would be okay ... with "nearly-inband" I
think of a second port the user can connect to to send binary data. E.g.
something like PASS. - I think if you can establish one TCP/IP
connection you can establish a second one two as long as it's always
established in the same directon (the problem with NAT and ftp is that
the second connection is originated by the server instead of by the client).
But sending big amounts of (binary) data through the server should
always be an exception, e.g. because the client is behind a NAT router.
Small amounts of binary data can also be sent inband the XML stream.
Fon: +49-700 77007770 http://matthias-wimmer.de/
Fax: +49-89 312 88654 jabber://email@example.com
More information about the Standards