Re: LWSP are ambiguous in ical-05

[Jerome Abela <Jerome.Abela@xxxxxx>]

On Dec 15, Matt Hainje wrote:
> I believe Jerome's concern is [...]

I sent 2 messages, and Philippe correctly answered one (about folding),
but quoted the other one (about LWSP).

>     SUMMARY: This is the summary.
> 
> Now, what's the SUMMARY value?
>  * "This is the summary." -or-
>  * " This is the summary."

That's it.

It was already discussed at Memphis, and the minutes say that
"non-significant LWSP-Characters should be restricted everywhere".
But the -05 grammar still includes LWSP-Characters. So my question
is: did I miss another discussion about this point ?

> It seems that we should only be defining type names, parameter names,
> parameter values, value elements, and any further restrictions, and the
> application of the grammar in 4.1.1 should then be applied.  This would
> simplify the structure of the iCalendar document significantly.

I agree.  Here is an example:

     contentline = name *(";" [WSP] param ) ":" [WSP] value-list CRLF
     [...]
     attach     = ("ATTACH" ":" [WSP] uri  CRLF) /
                  ("ATTACH" ";" [WSP] "ENCODING" "=" "b" ";" [WSP]
                  "VALUE" "=" "text" ":" [WSP] folded-b

could become:

     contentline = name *(";" [WSP] param ) ":" [WSP] value-list CRLF
     [...]
     The value of this property MAY be an uri or inline content
     information.

This gives the same meaning, without enforcing a more restricted grammar
than the liberal one suggested in 4.1.1.

Grammar would not be restricted anymore to only the few cases we may
have thought about: what about inline text (not binary encoded) ? What
about commas in uri ? What about a different charset for the inline text ?
None of them are legal with the Attachment definition of the -05 grammar.


Jerome.