From: Clark C . E. <cc...@cl...> - 2001-05-22 23:29:45
|
On Tue, May 22, 2001 at 03:23:14PM -0700, Brian Ingerson wrote: | BTW, What's the current syntax for refs and classes? My only input at | this point is that they both must _precede_ the value. That's because a | scalar no longer has a marker symbol. It would be ugly to put them | after, say, a hundred line string. Right. | Candidates: | | key : &0001 #class1 @ | #class2 "String Value" | &0003 % I think this was Oren's suggestion, and this is what I was intending to document. Thoughts? | key : (&0001 #class1) @ | (#class2) "String Value" | (&0003) % | | or | | key : &0001 (org.yaml.class1) @ | (org.yaml.class2) "String Value" | &0003 % | or key : &(0001) #(org.yaml.class) @ #(org.yaml.class2) "String Value" &(0003) % I really don't see how this is much different from Oren's syntax... other than it is different. | Or something else along this structure. The good thing about using | parens (at least for the class) is that we could treat it like a | comment. The application could use it as it saw fit. I think Oren | mentioned this? | | Parens might be a good candidate for your "comment" syntax in general. | Or at least for inline comments. I'm generally against comments since they bring up the awful question: "Are comments in the information model?" If they are, then they really arn't comments, and if they arn't, then they won't round trip and Preserving comments is useful. Thus comments are a bag of worms that most people find useful, but causes problems. So... I just wanted to ask to see if anyone _absolutely_ needed comments. If not, I'm willing to nix them. On the other hand, if we did allow comments, I was thinking that they could be comment nodes... and thus would have their own indicator. Since we have reserved several indicators, this makes sense. Clark |