- From: Rick Jelliffe <ricko@allette.com.au>
- Date: Mon, 23 Jul 2001 12:35:58 +0800
- To: <www-xml-blueberry-comments@w3.org>
- Cc: <xml-dev@lists.xml.org>
From: "John Cowan" <cowan@mercury.ccil.org> > > Under my proposal a Blueberry declaration is legal if and only > > if one or more Blueberry characters is used in an XML name somewhere in > > the document. Thus adding a single processing instruction whose target > > contained a Blueberry character either before or after the root element > > would make the document Blueberry legal. > > Well, I certainly have no problem with this idea. I certainly do. It goes against the fundamental principle of labelling. FrameMaker's MIF used a similar system to this, in a way: for language you type the country name in the native characters (e.g. idographic NI-HON for Japan) and then the MIF reader will know which encoding is being used (i.e., shift JIS or EUC I suppose) because it knows the two code sequences. But we didn't go that way in XML: instead, after the most minimal reliance on signatures (for determining code-point size and ASCII/EBCDIC family) it uses explicit labelling in text: markup. The advantage of this is that there is as much chance as possible that the document's character set can be known by using a standard hex dump, and usually using the system's standard text editor or "type" or "more" commands. I think it would be a great step backwards to have any system based on signatures, when we have managed to reduce their use to the minimum currently. Markup (i.e. ASCII declarations) is better. Cheers Rick Jelliffe
Received on Sunday, 22 July 2001 22:31:43 UTC