Re: [Yaml-core] Small Issues

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

On 15/11/01 10:35 +0200, Oren Ben-Kiki wrote:
> - Using an empty type string (just a '!') should force implicit transfer
> method detection on any scalar:
> 
> integer: ! "12"

Cool! But... 
Do we *need* this? I think overloading the syntax at this stage is a
mistake. We might want to reserve this for something else. Let's
consider this as a feature for YAML 1.1.

> - It would be more consistent to change '|-' to '||':
> 
> \  -     folded, not escaped
> \\ -     folded,     escaped
> |  - not folded, not chomped
> || - not folded,     chmoped

Well, OK. It's such an oddball case that I don't mind having an oddball 
syntax like |- to represent it. By the same terms I can't really argue 
much about using *any* syntax for it. So go ahead.

BTW, I don't remember discussing any chmop properties in the past.

> - Brain also wanted a leading '---' in the log file example. The very first
> '---' one is optional...

Well I think this Brain fellow is probably right. (Fitting name, that) 
The original proposal was that the first separator was optional only if 
the stream had one document. Otherwise, how do you know what the
separator is?

---

One more point. I want to ammend the rule of duplicate keys. 

   "When a Loader program encounters a duplicate key, it should warn and 
   then use the latter value."

This will just be too potentially useful. I think the warning is enough. 
Perl doesn't even warn on duplicate keys. It's a common idiom. I don't
want to dictate how people must use YAML.

BTW, a *Parser* should neither detect, nor care about duplicate keys.

Cheers, Brian