From: T. O. <tra...@ru...> - 2004-09-09 20:38:43
|
On Thursday 09 September 2004 03:50 pm, Oren Ben-Kiki wrote: > "!" is simply something else ("I have no tag"). Obviously %TAG doesn't > affect nodes without a tag. Hence %TAG doesn't affect "!". Does help. But that def. is too broad per my point out that it cause a contradiction. (a) "23" same as ! "23" (b) ! "23" means 23 hence "23" same as 23 ? No. So saying ! means "I have no tag" is not precise. In (b), ! is being used as a special operator to convert ?string to ?plain. This is different. This is blah (livable but blah). I like Sean's idea better: dealing with non-plain/escaping in resolution phase. Thinking of scalars in this way had not previously been the general concensus. Currently - | string - > string - "string" are all the same thing, just with different formatting. But now the will be considered ?string. But in fact the first two really don't make much sense as ?string cause they HAVE to end up as !!str if the formating is to be meaningful. Moreover, - string is really the same deal. Its a "string" with special formating --plain. So now you've added context: - Nodes with specified tags are loaded by-tag. - Nodes with unspecified tags are loaded by-content ... and context (path to the node). you might as well take the next step and add ... and scalar style Then, the spec will just have the three variants ?mapping, ?sequence and ?scalar. Then we move !!str to repository and define canonical forms for | > " " and so forth. -- T. |