Yow, I had spaced that sort of approach, thanks.
BTW, we are not using direct SCP because of some concerns we have with it.
Basically, you can probably crash any server out there which speaks it by
sending a packet with the maximum length (nigh 4 gigabyte). Since most
implementations I can think of for handling SCP packets would chunk them
on the fly (since the idea behind using packets of this sort is to break
data up into smaller chunks) and build a stream from them, nobody would
bother being concerned with super long packets. As it stands now I don't
know of a single normal computer which could handle a 4 gig packet even
IF it included swap space.
In the future computers may be able to control that, but assuming the
speed with which protocols are superseeded by newer better and faster
protocols we felt it was best to restrict the data length field to
< why poor?
Because I had not considered simply prefixing another scheme as you
did. I was thinking along the same lines as netscape's http's' protocol,
< I read the cold and ice cold pages. How is the un-named client coming on?
Quite well, Jeff Bellegarde is developing it. He is incorporating many
portions of Grail, which is speeding development a bit.