Re: [calsify] SKIP was Re: AD review of draft-ietf-calext-rscale-03

[Ken Murchison <[e-mail]>]

On 02/03/2015 03:08 PM, Marten Gajda wrote:
> @Ken: Sorry for the double post, I hit the wrong reply button
>
> Am 03.02.2015 um 19:13 schrieb Ken Murchison:
>> On 02/03/2015 12:58 PM, Ken Murchison wrote:
>>> On 02/03/2015 12:51 PM, Cyrus Daboo wrote:
>>>> Hi Ken,
>>>>
>>>> --On February 3, 2015 at 12:38:52 PM -0500 Ken Murchison 
>>>> <[e-mail]> wrote:
>>>>
>>>>> I believe we can get by with just a single SKIP and your previous 
>>>>> rules
>>>>> to handle both cases above (using Hebrew leap month):
>>>>>
>>>>> RRULE:FREQ=YEARLY;RSCALE=HEBREW;SKIP=FORWARD;BYMONTH=5L;BYMONTHDAY=30
>>>>>
>>>>> Would yield 1 Nisan (skipping forward both a month and day)
>>>>>
>>>>>
>>>>> RRULE:FREQ=YEARLY;RSCALE=HEBREW;SKIP=FORWARD;BYMONTH=5L;BYMONTHDAY=-1
>>>>>
>>>>> Would yield 29 Adar (skipping forward just a month)
>>>>>
>>>>
>>>> Actually I am not convinced about that since the spec says that 
>>>> SKIP is applied AFTER all other rule parts are processed (except 
>>>> for BYSETPOS, COUNT and UNTIL). So the BYMONTHDAY=-1 refers to the 
>>>> last day of the invalid month, 30 Adar I, which then becomes the 
>>>> invalid date 30 Adar after the month skip and then 1 Nisan after 
>>>> the day skip.
>>>
>>> Hmm, I think you're right based on the current wording.  I don't 
>>> recall exactly how we decided on that wording, but perhaps it might 
>>> needs to be adjusted.
>>
>> Looking at RFC 5545:
>>
>>       If multiple BYxxx rule parts are specified, then after evaluating
>>       the specified FREQ and INTERVAL rule parts, the BYxxx rule parts
>>       are applied to the current set of evaluated occurrences in the
>>       following order: BYMONTH, BYWEEKNO, BYYEARDAY, BYMONTHDAY, BYDAY,
>>       BYHOUR, BYMINUTE, BYSECOND and BYSETPOS; then COUNT and UNTIL are
>>       evaluated.
>>
>>
>> I'm wondering if we shouldn't have SKIP applied both immediately 
>> following BYMONTH and BYMONTHDAY evaluation.  I don't think we want 
>> the other BYxxx rule parts to be operating on invalid dates at any 
>> point.  Thoughts?
>>
> The only reason I can come up with right now, is that a rule with 
> RSCALE & SKIP=OMIT could have different results than the same rule 
> without RSCALE and SKIP.
>
> example:
>
> DTSTART;VALUE=DATE:20120229
> RRULE:FREQ=YEARLY;BYMONTH=2;BYDAY=SU
>
> -> all Sundays in February, every year
>
> DTSTART;VALUE=DATE:20120229
> RRULE:FREQ=YEARLY;RSCALE=GREGORIAN;BYMONTH=2;BYDAY=SU;SKIP=OMIT
>
> -> all Sundays in February, but only in leap years, since SKIP would 
> filter Feb. 29th in non-leap years before BYDAY is evaluated

I think both of these come out the same (and my libical computes the 
same) as there is no BYMONTHDAY or anything else forcing the expansion 
onto Feb 29 in a non-leap year.  In this case the DTSTART just gives us 
the first instance (which in this case doesn't fall on a Sunday so isn't 
included).  After that, the BYMONTH+BYDAY is what really drives the 
expansion.

-- 
Kenneth Murchison
Principal Systems Software Engineer
Carnegie Mellon University