IPP Mail Archive: Re: IPP> Chunked POST

Re: IPP> Chunked POST

Carl Kugler (kugler@us.ibm.com)
17 Dec 1998 16:54:45 -0000

Michael Sweet
> Carl Kugler wrote:
> >
> > >
> > > How many IPP implementers are using an existing HTTP server?
> >
> > How many platforms don't already have one?
> Well, that's the problem, isn't it? I know of at least three dozen
> commercial and freeware HTTP servers for UNIX and Windows, each with
> its own idiosyncracies. As a software vendor, I can't expect that
> 1) they'll have a specific brand of HTTP server that I've tested on,
> and 2) they'll even have a HTTP server loaded and running on a
> system (many government sites *cannot* have them because of new,
> more strict network policies).
> Soo, we've implemented our own small HTTP/1.1 server within our
> new printing system (CUPS). It makes the design of the printing
> system simpler (no double layer of communication) and we aren't
> dependent on any one vendor or configuration.
I think that's a good solution for the short term, but in the long run I think the httpd should be part of the platform infrastructure, not something that has to be written and installed again for each HTTP application. Call me a dreamer; I haven't given up on the Grail of modular software and code reuse.

> Embedded products (read: printers) also generally don't have HTTP
> servers (there are exceptions, like HP and Tektronix for example).
> The implmentation issues involved with network printers can also
> prevent you from using existing source code...
I just think it's likely that vendors that deploy IPP are likely to have other uses for an http server in the Printer. Especially in an embedded environment, it's a shame to have to put in redundant implementations.

> > ...
> > Unfortunately, due to the CGI restriction, most http servers seem to
> > be designed to discard the message-body of an HTTP/1.1 POST without
> > Content-Length, even if the target is not CGI but, say, a servlet.
> > Right now, I'm just trying to confirm whether or not this is broken
> > behavior.
> From the standpoint of HTTP/1.1, it certainly is. I think IPP is
> probably the first (of many) HTTP applications that will expose these
> bugs...
> --

See the original message at http://www.egroups.com/list/ipp/?start=4999