>
> The problem is already solved if the roles are reversed:
>
> Customer: POSTs URI of file to Vendor
> Vendor: GETs the file from Customer
>
> Yeah, yeah... easier said than done. So is making HTTP stateful.
>
To elaborate... I've never coded a browser. However, almost every webserver project I've ever been involved with since circa '96 has used libcurl to fire off a GET to some other server, in response to a GET from a browser, so all my servers are also clients. Why can't that GET go right back through the tunnel for a file:// URI provided in a POST?
I don't have an ivory tower to climb down from, just a skunkworks to crawl out of (ask my dogs, lol). After I got my B.S. in Chemistry, I went skiing for a few years, before opening my ISP / Web shop. When I worked the top lift shack, I would read the foundational RFC's I'd printed off (dot matrix tractor feed) the night before, as they downloaded from Compuserve at 2400 baud (on a Packard Bell 486).
Never met nor corresponded with Jon Postel, but I couldn't have gotten a better education even if Internet *was* taught when I was at University, or even after when Evi Nemeth was my friend, neighbor (weekends, while ashore), and mentor -- although I was not a munchkin.
[link] https://en.wikipedia.org/wiki/Jon_Postel
[link] https://en.wikipedia.org/wiki/Evi_Nemethhttps://www.admin.com/samples/EviTribute.pdfhttps://en.wikipedia.org/wiki/Evi_Nemeth
[link] https://www.admin.com/samples/EviTribute.pdf
We all stand on the shoulders of giants. If I live another year I'll be older than Jon Postel or Jerry Garcia? Man that makes me feel old. But I know both of the departed souls I linked to above, are more than happy that I try to build on their legacies. I consider Jon's 200+ RFCs and Evi's book, required reading. Also Roy's dissertation, when it comes to HTTP.
-Eric