Venn diagram in -defs-06

John C Klensin klensin at jck.com
Tue Mar 3 18:19:00 CET 2009



--On Tuesday, March 03, 2009 11:29 -0500 Andrew Sullivan
<ajs at shinkuro.com> wrote:

> Dear colleagues,
> 
> It appears to me that the Venn diagram at Fig 1 in -defs-06
> has a mistake in it.  In it, R-LDH LABELS are contained inside
> the box of NR-LDH LABELS.  I think that's wrong.  As I read
> the definitions, R-LDH and NR-LDH are two completely disjunct
> sets.  (If you have trouble reading the diagram in the draft
> because of the way it's been laid out across the pages by the
> rendering tool, you can find it also in
> http://www.alvestrand.no/pipermail/idna-update/2009-February/0
> 03463.html.)
> 
> The other possibility is that there's something wrong with the
> diagram.  As I look at the diagram, there does appear to be
> something wrong with it.  If we work outwards in the top
> boxes, we have A-labels and Invalid A-labels in the innermost,
> then XN LABELS, then R-LDH LABELS, then NR-LDH LABELS, then
> LDH-LABEL, which means that the outer ASCII-LABEL label is
> superfluous.  That can't be right.  So if I work in, I read
> ASCII-LABEL as the marker for the outer box.  LDH-LABEL is
> then the marker for the top large box (that also makes sense,
> because otherwise that box is both LDH-LABEL and
> NON-LDH-LABEL, and I'm pretty sure Venn diagrams don't allow
> A&~A in the same box ;-) .  Since A-labels, Invalid A-labels,
> XN LABELS all obviously mark the three innermost boxes, we're
> stuck with two markers (R-LDH and NR-LDH) for one box.
> 
> I attach a PNG file (I dunno if it'll make it to the list --
> if it's filtered I'll try something else) that I think is the
> diagram actually wanted.  I am nowhere near good enough a
> warlord to do it in ASCII-art.  I didn't bother labelling the
> subclasses of Non-LDH-labels, since everything in that class
> is automatically out of scope for us anyway.
> 
> I think this needs fixing up, or else it needs to be removed
> from defs.

Andrew,

Thanks.   I did receive the picture and will try again on the
draft.    I suggest that people keep both defs-06 and the
(forthcoming) defs-07 for comparison.   

I agree that, if we can't get the picture right, we should pull
it. But enough people have said that it is helpful, and no one
so far has asserted that it is not, that I hope we can get it
right.

     john



More information about the Idna-update mailing list