[5705] in www-talk@info.cern.ch
Re: holding connections open: a modest proposal
daemon@ATHENA.MIT.EDU (Rick Troth)
Mon Sep 19 11:54:31 1994
Date: Mon, 19 Sep 1994 17:45:28 +0200
Errors-To: listmaster@www0.cern.ch
Errors-To: listmaster@www0.cern.ch
Reply-To: TROTH@UA1VM.UA.EDU
From: Rick Troth <TROTH@UA1VM.UA.EDU>
To: Multiple recipients of list <www-talk@www0.cern.ch>
>Stealing from PostScript, where the same situation applies (with the Bounding
>Box) when programs are generating the file and writing as they go:
>
>[stuff]
>Content-Transfer-Encoding: binary
>Content-Length: (atend)
>[lots of stuff]
>Content-Length: 12345678
That's an interesting theft.
>Yes I know this raises other problems, it was just a suggestion. Does the
>program parsing MIME have to know the length at the beginning? Is it OK to know
>it later on, check against how many bytes were recieved, and detect it has got
>them all?
Iff you can somehow switch back to "header interpretation" mode.
You've got to use some kind of boundary or delimiter. Otherwise the
"Content-Length: 12345678" will be seen as just more of the data. :-(
I addressed this problem when creating something I call
Sender Initiated File Transfer. I didn't want to use a second stream,
but couldn't always know ahead of time how big the file would be,
so I sent things in "segments" (bursts), x number of bytes at a time.
But that's not HTTP. :-(
>--
>Chris
--
Rick Troth, <rmtroth@aol.com>, <troth@ua1vm.ua.edu>, Houston, Texas, USA
