Venn diagram in -defs-06
John C Klensin
klensin at jck.com
Tue Mar 3 20:07:08 CET 2009
--On Tuesday, March 03, 2009 10:40 -0800 Erik van der Poel
<erikv at google.com> wrote:
> Actually, binary and bit-string values might sometimes consist
> entirely of octets less than 128, so the top-level split
> should be:
> text vs non-text
> Then non-text can be split into binary vs bit-string, and text
> can be split into ASCII vs Non-ASCII.
Three observations from your frustrated editor (and apologies in
advance if the frustration comes through too much):
(1) Bit-string values are a different label type, so
they aren't part of a top-level split and their octet
values are substantially irrelevant. No disrespect
intended, but it would really be helpful if people
engaging in this discussion spent a bit of time
understanding DNS fundamentals.
(2) These pictures are tricky to construct and trickier
yet to fit neatly within the 69 character by 49 line
constraint of the RFC format. I've had several "can't
you stop Figure 1 from splitting across page boundaries"
comments. The answer is that, at present, the "figure"
includes the comment that precedes it and the notes that
follow, which causes it to exceed that limit, so "no".
I can make it fit by pulling the notes into a separate
text block and will do that when (and if) we are
satisfied about the content being right. However, I'll
have to force things to stay together when I do that
with manual page breaks. That is error-prone and likely
to be fouled up by further editing, so my inclination to
do it while we are still making changes is very low.
Incidentally, the reason for splitting off Figure 2 from
Figure 1 was related to the formatting issue, not a
(3) Several weeks ago, after Defs-04 was posted, there
there were multiple comments on the list that indicated
that the picture --not just the idea of a picture-- was
helpful. I hope the changes and corrections since
(including the one suggested by Andrew) helped clarify
it further. But the more we make it complicated by
hair-splitting, the less useful it becomes. If the
consensus now is that the picture cannot be represented
in a satisfactory way unless far more information is
inserted than I can fit, I can certainly pull it back
So, to be very explicit about this...
(i) Do you (and others) believe that the pictures have outlived
their usefulness and should be pulled out?
(ii) If not, and given the nature of "bit-string" discussed
above, do you think it is useful to cut the picture a different
way? And, if so, can you supply an ASCII text file that
contains exactly what you want with a caption and image area
that occupies no more than a 69 character by 49 line block?
More information about the Idna-update