I do. Of course I also think a web service interface on just about everything makes sense.

I hope that if iCalendar validation advances in 2009, it’ll take that approach.

I’d also love to see it follow the approach taken at feedvalidator.org w/respect to test cases.

http://feedvalidator.org/testcases/

Although the validator itself is written in Python, the tests are purely declarative and completely decoupled from the implementation. A different validator could work off the same tests, and multiple validators could serve as checks on one another.

What’s more, if that happened, the ensuing conversation would doubtless improve the breadth and quality of the (already excellent) test suite.

I would then be able to do more with it than I have been lately — because I can’t recommend something that makes warnings look like errors, esp when they’re so much a matter of taste and don’t reflect anything the spec says or doesn’t say.

I was looking for a recent thread that mentioned the validator and this is the one I found — I thought you might agree that a web service interface for the validator makes sense.

Hope all is well and that you’re having a pleasant and healthy 2009!

Although, as we see above, their former non-validity didn’t prevent three iCalendar libraries from parsing the feeds.

“I had to use a regular expression and pytz to take the timezone information and transform the start and end times.”

And therein lies the dilemma. If those changes are required, a robust validator should prescribe them.

“Also, when looking around my community (New Boston, NH) to see if I could create an aggregated calendar, like elmcity.info, I found nary an iCalendar enabled calendar!”

This, of course, is the ultimate purpose of my project. ICS feeds want to be as superabundant as RSS feeds. They aren’t, and for no better reason than that it just doesn’t occur to people to publish them.

Changing that is an exercise in politics more than in software development. Jim Groom’s going to try to make Fredericksburg VA a sister city to Keene w/respect to this idea:

http://bavatuesdays.com/a-calendar-year/

Maybe you can do the same for your region of NH?

I updated the chart above to include vObject. As you can see, it seems to like everything except the two Drupal-produced feeds.

“It’d be great to collect a snapshot of failing icalendar files and catalog them just so implementers in all languages could see which ones they fail on, how!”

Absolutely. That’s another benefit of maintaining the feed registry in an open, queryable location:

http://del.icio.us/elmcity/trusted+ics+feed

“I think the folks at CalConnect came up with test files like this at some point, but I don’t think they were made public.”

In any case, the stuff that comes up in the wild is what’s most interesting to me.

“If you aren’t already planning on it, you might try to attend the CalConnect at Microsoft’s Redmond campus, http://www.calconnect.org/calconnect14.shtml”

Hmm. Perhaps. Although I don’t think that the use case I’m pursuing here is of much interest to that community.

I’m using iCalendar.py on my GAE project (www.mashical.com) and have felt the pain of ics formats. For example, Yahoo sports calendars use TZID in the DTSTART and DTEND, which iCalendar.py doesn’t handle natively. I had to use a regular expression and pytz to take the timezone information and transform the start and end times.

Also, when looking around my community (New Boston, NH) to see if I could create an aggregated calendar, like elmcity.info, I found nary an iCalendar enabled calendar!

(as an aside, I’d be fascinated to see which of your feeds vobject fails to parse).

The validation scaffolding is still in there, though, and it wouldn’t be particularly hard to improve.

It’d be great to collect a snapshot of failing icalendar files and catalog them just so implementers in all languages could see which ones they fail on, how! I think the folks at CalConnect came up with test files like this at some point, but I don’t think they were made public.

I keep meaning to make a GAE icalendar validator using vobject, maybe one of these weekends.

P.S. If you aren’t already planning on it, you might try to attend the CalConnect at Microsoft’s Redmond campus, http://www.calconnect.org/calconnect14.shtml