[IPFIX] [Errata Verified] RFC7011 (7875)

[RFC Errata System <[e-mail]>]

The following errata report has been verified for RFC7011,
"Specification of the IP Flow Information Export (IPFIX) Protocol for the Exchange of Flow Information". 

--------------------------------------
You may review the report below and at:
https://www.rfc-editor.org/errata/eid7875

--------------------------------------
Status: Verified
Type: Technical

Reported by: Katherine Prevost <[e-mail]>
Date Reported: 2024-03-28
Verified by: Warren Kumari (Ops AD) (IESG)

Section: 6.2

Original Text
-------------
   Reduced-size encoding MAY be applied to the following integer types:
   unsigned64, signed64, unsigned32, signed32, unsigned16, and signed16.
   The signed versus unsigned property of the reported value MUST be
   preserved.  The reduction in size can be to any number of octets
   smaller than the original type if the data value still fits, i.e., so
   that only leading zeroes are dropped.  For example, an unsigned64 can
   be reduced in size to 7, 6, 5, 4, 3, 2, or 1 octet(s).

Corrected Text
--------------
   Reduced-size encoding MAY be applied to the following integer types:
   unsigned64, signed64, unsigned32, signed32, unsigned16, and signed16.
   The signed versus unsigned property of the reported value MUST be
   preserved. The reduction in size can be to any number of octets
   smaller than the original type if the data value still fits, i.e., so
   that only leading zeroes are dropped (Or, in the case of negative
   signed values, so that only leading ones are dropped). For example,
   an unsigned64 can be reduced in size to 7, 6, 5, 4, 3, 2, or 1 octet(s).

Notes
-----
As written, the specification suggests that integer values may only be encoded using a reduced-size encoding if they have a run of zeroes at the beginning. However, the specification also describes being able to use reduced-size encoding for signed integer values—and only non-negative values of signed integers have a representation that begins with zeroes.

Either the text should clarify that negative values may never be expressed using a reduced-size encoding (which is what the strictest reading of the current text would suggest), or it should specify that leading ones may also be dropped for signed values, which makes it clear that they should be interpreted via sign extension of the high bit.

A quick survey of existing IPFIX implementations suggests that those which decode reduced-size encoding of integers at all produce the semantic equivalent of sign extension when they encounter a reduced-size encoding of a signed integer value.

-- 
WK: Thread (for posterity): https://mailarchive.ietf.org/arch/msg/ipfix/sBsLy-XYH8sQCEjnPSGoEKsRL9Y/


--------------------------------------
RFC7011 (draft-ietf-ipfix-protocol-rfc5101bis-10)
--------------------------------------
Title               : Specification of the IP Flow Information Export (IPFIX) Protocol for the Exchange of Flow Information
Publication Date    : September 2013
Author(s)           : B. Claise, Ed., B. Trammell, Ed., P. Aitken
Category            : INTERNET STANDARD
Source              : IP Flow Information Export
Stream              : IETF
Verifying Party     : IESG