From: Clark C. E. <cc...@cl...> - 2004-09-13 07:07:03
|
On Mon, Sep 13, 2004 at 02:54:23AM -0400, trans. (T. Onoma) wrote: | > The previous two options are: | > | > - We report four ?tags to the application, ?mapping, ?sequence, | > ?scalar-plain, ?scalar-other. | | Either this with: ?map, ?seq, ?sym, ?str Well, in any case, we still need to provide the quoted style for these ?sym objects. So, some way to flag the quoted style that it is a ?sym would be needed. "quoted" -> ?str ! "quoted" -> ?sym Its icky, but the only clear option I see. If we just make ?sym a ?str with a tag of ! it is the same as above, and we don't need the extra ?thingy. So I still prefer the latter option, 23 = ! "23" ;) | No, better. Have _four kinds_. the all tags can be NULL. The | kinds are mapping, sequence, string, symbol (the last two | being the two fundamental types of scalars) I also prefer NULL; but I don't see a fundamental difference in 'string' and 'symbol', they can both be typed !!int correct? | > - We report three NULL tags to the application for the untagged | > mapping, sequence, and non-plain scalar. The plain scalar | > is reported using the ! tag. | > | > The primary advantage of the latter one is that you can choose to | > only use double-quoted strings (necessary for my canonical form use | > case), since these two are equivalent: | > | > - 23 | > - ! "23" | > | > So, in effect, a plain scalar becomes syntax-short-hand for using a | > ! private type. Of course, this will cause Oren to have cats since | > he wants the untagged nodes to be NULL. |