- From: Brian Behlendorf <brian@organic.com>
- Date: Mon, 13 Nov 1995 15:15:17 -0800 (PST)
- To: Chuck Shotton <cshotton@biap.com>
- Cc: Benjamin Franz <snowhare@netimages.com>, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
On Mon, 13 Nov 1995, Chuck Shotton wrote: > I want to see some > constructive discussion about why a CGI-based implementation of byte ranges > is unacceptable. To use your Xmodem/Zmodem analogy, it's like suggesting that download parameters be tied into a name of an object to be transferred. Using a CGI script to deliver parts of files is definitely not "wrong". However, writing a PDF viewer which presumed any server serving PDF files to also have a "/cgi-bin/pdfsplitter" CGI script or something in its namespace is going beyond the bounds of what a transport or communications protocol should define. The point of trying to decide on a "standard" for this is not just so that server authors can agree to implement common support - proxy servers need to know that a request for part of a thing can be immediately fulfilled if the whole thing is in the cache. This is presumably why John and Ari brought it to the table in the first place - so that Ari can support this kind of functionality in the netscape proxy. A CGI-based implementation makes this impossible, unless you suggest that the proxy recognize and intercept requests to /cgi-bin/pdfsplitter or whatever it is called. I believe what you are really asking for is a way for the client and server to have something of a conversation about how to access sub-parts of content - i.e., frames of an movie, time ranges of audio, a page out of a PDF file. Obviously byteranges are inadequate for these kind of application-level queries. Let's <em>separate</em> this from the byterange proposal for the time being - there are definitely better answers to this problem, Hytime is/was one solution, and I think it's ill-advised to hold everything up until it's solved. Focus on byte ranges as being specifically for transmission recovery for incomplete files - the fact that some companies will be using them to access individual pages out of a PDF file should just be considered an abuse. > I would like to see some rationale behind why the URL standard needs to be > changed to support an application-specific form of query for a specific > subset of data types. You're right, it doesn't, I'm with you there. > I'd like to see a well-reasoned discussion on your > part about the valid concerns that server implementors have regarding the > need to re-render entire documents to serve you a byte range for low-value > reasons like interrupted file transfers, etc. and why you think it is OK to > ignore these concerns. The algorithm I posted yesterday allows for servers to return full objects even when only a part is requested - the choice is up to them, since that point may be hard to compute or dynamic or whatever. So, I would presume in your case the server would just always return full objects. No sweat - that's what the current netscape implementation presumes anyways. Recognizing byte ranges is just a window for optimization. If you don't want to do it, fine, you don't lose. > You seem to want to trivialize the function of the Web and associated > applications to the lowest common denominator of moving the contents of a > file from point A to point B, a byte at a time. This is an absurdly > low-level of abstraction. It's like talking about building a house by > aligning protein and sugar molecules so that they form 2x4s, lining up iron > atoms into nails, etc. I hear Eric Drexler is making a comeback.... Brian --=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-- brian@organic.com brian@hyperreal.com http://www.[hyperreal,organic].com/
Received on Monday, 13 November 1995 16:08:57 UTC