Re: [Yaml-core] Pesky question

[Brian I. <in...@tt...>]

On 03/12/01 11:40 +0200, Oren Ben-Kiki wrote:
> - Alpha (in transfer/directive "words", for implicit !str text, in anchors).
> Here I see a problem...
> 
> For transfer/directive names "words" I think we should stick with ASCII
> alpha characters. I think that's what IANA and DNS require today anyway.

OK

> For detecting implicit !str text - I don't know.
> 
> - We could say that any 'L&' class Unicode character indicates !str. This
> means every YAML implementation would require the Unicode tables (which are
> not too big but ugly and prone to change). Also I'm not certain that 'L&' is
> the right class... We'd need a Unicode wizard to specify the list of
> relevant classes - do any of you guys have access to one?
> 
> - We could say that any character > 127 indicates !str. This simplifies the
> implementation considerably, at the cost of disallowing implicit transfer to
> be indicated by non-ASCII, non-alpha characters. Unicode has *a lot* of
> these beasts.
> 
> I tend to the second option. It means that there would be very few implicit
> transfer types, but I think that's a good thing.

Sounds good.

> For anchors, I suggest we say "anchor ::= non_space_in_line+" - allow any
> string whatsoever, thereby avoid the issue altogether. I don't see the gain
> in restricting it to "one word".

OK. BTW, I think we can omit the leading zero nonsense for anchors now.
Anchors are non_space_in_line strings period.

I'm trying to pick a good canonical startig anchor for the Store function.
Prescod suggested "&ID0001". What do you guys think?

> 
> - Simply allow every character >= #x20.
> 
> Again my view is that allowing every character >= #x20 is probably enough.

Fine by me.

I'm sure Clark will be more opinionated on these issues though.

Cheers, Brian