Textual Representation of IP Flow Information Export (IPFIX) Abstract Data Types
draft-ietf-ipfix-text-adt-10

Should this be PS instead of Informational?

-- Section 4.1 --

   octetarray = 1*(hex-octet [WSP])

I'll note that this allows an octetarray to have trailing WSP.  If that's intended, that's fine, but it does differ from what the text says.  If you want to fix it, this will work:

NEW
   octetarray = hex-octet *([WSP] hex-octet)
END

-- Section 4.3 --
Please fix the table entry for "signed64" so that the minimum value is all on one line.

-- Section 4.4 --
In the ABNF for "naninf", it's not strictly required but it would help readability to put parentheses around << sign "inf" >>.

-- Section 4.7 --
Last paragraph: nicely dodged.  :-)

I'd just note that some of the considerations expressed in section 4.7 do seem like security considerations given the relative dangers associated with string processing.

I question along with Alia: Why is this Informational? Seems like these things will eventually be protocol elements.

4.4:

( "e" / "E" ) is not necessary; in ABNF, any character enclosed in quotes is case insensitive.

The ABNF as written allows the mantissa to terminate with a decimal. So you could end up with:

   1234.E56

Is that legit? Did you instead mean:

   right-decimal = "." 1*DIGIT

?

4.8: I do wish you didn't reuse "partial-time" when it differs from the definition in 3339. How about "whole-time" or "integer-time"?

4.8-4.10: If you are importing the ABNF from somewhere else, it might be better to just do so by reference instead of reproducing the text here.

Is there any particular reason to insist that white space in octet arrays be inserted at octet boundaries?  It seems like you could just have the ABNF be "octetarray = 1*(hex-octet [WSP])", and deal with odd-length arrays either by forbidding them in text or by zero-padding.

It could be helpful to clarify that the min/max values in Table 1 and Table 2 are decimal.

It seems odd to allow hex/binary for unsigned, but only decimal for signed.  Couldn't you just say "signed = [sign] unsigned"?

Nit: ".." at the end of Section 4.7.

Thanks for resolving my discuss with the additional security
considerations.