This is, I feel, a serious mistake. Duration can well be beyond 12 months. 
In IT forklore we have the 18 month duration in a number of "Moore's law" 
variants--- used by industry leader Intel in their own roadmaps and stategic 
planning goals to self-fulfill.
The urge to canonization and normalization has, I think, here resulted in a  
shot in the foot..
Please also note that P12M and P1Y are not the same even if, by definition, 
1 year is 12 months, yes.. its that mantra about precision again :-)

One should think about what durations expressed in units of precision finer 
than the base mean. For example:  2001/P2M
I suggest there is an equivalent expression using our post-fix "?" and "~" 
operators but will leave that discussion for another day...

I will quickly (saving my usual verbage) suggest that we should:
1) drop this limitation in months.
1) extend the units to duration to century, hours, minutes and decimal 
3) allow for positive decimal values rather than limit things to integers.
4) add a clause requesting that, whenever, possible some canonical 
normalized form be used. The question of what it is, I think, we should 
leave open at this time. P1.5Y versus P18M.. ?

On Wed, 11 May 2011 09:13:45 -0400, Ray Denenberg wrote
> From: [UTF-8?]SaaÅ¡ha Metsärantala
> > > why do we even need yearMonth, when month alone would be sufficient.
> > I should have clarified my aim with that. In one way, it would be
> > easier to only use month, but for long periods, I thought it would be
> > easier to read /P123Y than /P1476M. As you suggest, it rises the
> > question of c14n but this may be solved by a reformulation such as:
> > 
> > monthsDuration = oneThru11 "M"
> Ok, I've made this change.


Edward C. Zimmermann, NONMONOTONIC LAB
Umsatz-St-ID: DE130492967