[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Minimum IP Compression Algorithm for Interoperability
<David_Mason@nai.com> wrote:
> The problem was that RFC 2394 did not make it "_perfectly_
> clear that the header and checksum MUST NOT be used".
Agreed! 2394/1951 don't explicitly mention those facilities at all.
> Unfortunately both RFC 2394 and 1951 refer to the zlib
> source/documentation for use as a freely available
> deflate compliant package without mentioning the fact
> that the header/checksum should not be used. Since the
> header/checksum is the default mode of zlib and that not
> using the header/checksum feature is undocumented (and
> RFC 1950 mandates use of the checksum) this caused problems.
However, if folks stuck strictly with the bits/protocols specified in
the RFCs 2393, 2394, and 1951 one would not implement the
header/checksums. Its only when one looks further into the Zlib
references, uses Zlib, or digs up RFC 1950 (which isn't referenced as
an RFC in any of 2393, 2394, 1951) that one would implement the
additional header/checksum artifacts.
The specs are fine as is but again I'd support an "implementation hint"
in an appendix of 2394 to warn folks that wish to use off-the-shelf
deflate packages that they should be wary that it produce/accept
2394/1951 encodings only. Writing down any "undocumented" methods to
suppress the extra artifacts that folks have discovered into this same
appendix sounds like a helpful suggestion. I believe we've had this
suggestion since January, just no consensus yet to do it.
-- Marc --