Loading ...
Sorry, an error occurred while loading the content.

Reality of "Transfer-Encoding"?

Expand Messages
  • Karr, David
    Note that I m very new to exploration of the details of the HTTP spec, much less its implementation. I would assume that, like many protocol specifications
    Message 1 of 2 , Aug 1 8:55 AM
    • 0 Attachment
      Note that I'm very new to exploration of the details of the HTTP spec, much
      less its implementation.

      I would assume that, like many protocol specifications that are designed to
      meet a wide range of needs, there are areas of the HTTP specification which
      are less used, or rarely implemented, compared to other areas of the
      specification. I wonder if there is any understanding of what areas of the
      HTTP specification are essentially "moot", or are expected to not be ever
      used by real-world tools?

      In particular, I wonder about "Transfer-Encoding". Is this something that
      is really used? In addition, how relevant are "byte ranges", and even
      worse, byte ranges on entities requested with transfer-encodings?
    • Alex Rousskov
      ... Yes. AFAIK, several Squid-based projects use Transfer-Encoding to compress entities in transfer between cooperating intermediaries. I would imagine that
      Message 2 of 2 , Aug 1 10:07 AM
      • 0 Attachment
        On Wed, 1 Aug 2001, Karr, David wrote:

        > In particular, I wonder about "Transfer-Encoding". Is this
        > something that is really used?

        Yes. AFAIK, several Squid-based projects use Transfer-Encoding to
        compress entities in transfer between cooperating intermediaries. I
        would imagine that wireless and similar applications make use of
        Transfer- (hop-by-hop) and Content-Encoding (end-to-end).
        http://squid.sourceforge.net/projects.html#te

        I heard that new versions of ICAP require support for chunked transfer
        encoding.

        > In addition, how relevant are "byte ranges",

        Very relevant for handling large documents. For example, Acrobat
        Reader uses byte ranges to fetch (portions of) PDF files.

        There are lots of Web clients that use byte ranges to fetch large
        entities in small chunks (for reliability or whatever reasons). Web
        crawlers use that feature as well. Our Web server log is full of 206
        responses for source code/binaries downloads. I think those files get
        more range requests than "normal" requests!

        > and even worse, byte ranges on entities requested with
        > transfer-encodings?

        Not sure, but it seems to me that correct support for
        Transfer-Encoding and for ranges would yield correct support for
        ranges on transfer-encoded entities.

        $0.02,

        Alex.
      Your message has been successfully submitted and would be delivered to recipients shortly.