Re: [Orgmode] .ics export violates RFC2445

[Adam Spiers]

Carsten Dominik (address@hidden) wrote:
> On Apr 29, 2008, at 4:12 PM, Adam Spiers wrote:
> >Carsten Dominik (address@hidden) wrote:
> >>On  7Nov2007, at 9:56 PM, Adam Spiers wrote:
> >>>I use org-export-icalendar-combine-agenda-files to export my
> >>>appointments to an .ics file which I point korganizer at.
> >>>
> >>>I noticed ages ago that if I have an appointment with a comma in,
> >>>e.g.:
> >>>
> >>>** <2007-12-07 Fri 20:00> foo, bar
> >>>
> >>>korganizer always shows it as "bar" rather than "foo, bar".  But I
> >>>never got round to investigating whether it was a bug with the  
> >>>export or korganizer or something else ... until now :-)  I just took a  
> >>>quick look at the iCalendar spec, which is RFC2445, and discovered that  
> >>>the SUMMARY field is defined as follows
> >>>
> >>> summary    = "SUMMARY" summparam ":" text CRLF
> >>>
> >>>  -- from http://tools.ietf.org/html/rfc2445#section-4.8.1.12
> >>>
> >>>And the definition of 'text' in this context explicitly states that
> >>>several characters, including commas, need to be escaped with a
> >>>backslash:
> >>>
> >>> http://tools.ietf.org/html/rfc2445#section-4.3.11
> >>>
> >>>Sure enough, when I edited the .ics file and manually escaped the
> >>>comma, korganizer displayed the summary correctly.
> >>
> >>fixed, thanks
> >>
> >>- Carsten
> >
> >This appears to have regressed in some recent version ...
> 
> Yes, seems there was still a bug.  Fixed now.

Works great, thanks!

> >Also, it would be great if a UID field could be generated for each
> >event, perhaps by checksumming the contents of the event in some way.
> >The RFC says:
> >
> >  Conformance: The property MUST be specified in the "VEVENT",
> >  "VTODO", "VJOURNAL" or "VFREEBUSY" calendar components.
> >
> >    -- http://tools.ietf.org/html/rfc2445#section-4.8.4.7
> >
> >The checksum would ensure that the UID field only changes when the
> >event details check, which would be a first step towards helping
> >synchronisation systems.  I'm vaguely suspicious that the lack of UIDs
> >currently confuses Google Calendar too.
> 
> A UID may be good.  However I think changing the UID when changing the
> entry would be bad, because this would exactly *disable* synchronization.
> To synchronize, you must know which entries to compare, and this is only
> possible with a persistent UID.

Doh - you are right of course.

> I guess we could create one, but this UID would then have to be stored  
> in the entry, as a property.  Exporting to ical again must then re-use the  
> old uid each time.
> 
> My org-id.el in the contrib directory allows already to create unique
> identifiers, and it would be easy enough to include the domain to make  
> them truely unique, wordwide.
> 
> However, right now I am hesitating to force a property drawer onto  
> every entry that ever is exported to iCalendar.  But as an option, this might 
>  
> really be good and eventually allow true synchronization.

I understand your hesitation - as an option that sounds perfect.  I
forgot to mention before that gcaldaemon

  http://gcaldaemon.sourceforge.net/

which is a very nice piece of software indeed, requires the UID field.
If you provided that, then gcaldaemon would give us the solution to at
least unidirectional synchronisation with Google Calendar.  I have
experimented with simply scp'ing the generated .ics file to a private
location on my webserver and pointing Google Calendar at that private
URL, but there is no way to force Google Calendar to reload the data
after subsequent scp invocations.

> P.S. Adam, I emailed you twice privately in the last few weeks, but
> did not get any reply.

Hrm :-/  I have issues with private mail currently due to receiving
between 5,000 and 10,000 spam a day (not to mention being busy and
still having yet to perfect GTD ;-).  However I just replied to a
couple of recent ones - were they the ones you were referring to?