If that's the case, then why did Stefano's case fail? His input files caused Xalan to try to insert a single text node under the Document node, which was disallowed.
Mike Scott Boag/CAM/Lotus wrote: > > It doesn't need to. It can output directly to a stream. Xalan is in no > way DOM dependent for output. > > It determines if a stylesheet is meant to go to a text stream via: > <xsl:output method="text"/>, which is standard. > > If you just used <TEXTROOT>, you would still get XML (with escapes, etc.), > instead of text, which is what you want, I think. > > -scott > > > Mike Pogue > <[EMAIL PROTECTED] To: [EMAIL PROTECTED] > e.org> cc: (bcc: Scott Boag/CAM/Lotus) > Subject: Re: Xerces "feature" > prevents many useful operations > 12/17/99 > 04:11 PM > Please > respond to > xerces-dev > > > > I suppose if Xalan knew that the output was text, it could add something > like a <TEXTROOT> node, and then strip it off on output. Does this > work? > > Mike > > Tom Palmer wrote: > > > > > 2. Can a text node be a child of a document node? This is a DOM > > > level 1 question, and the answer is no, which is why the Xerces > > > DOM implementation was testing for and disallowing it. > > > > > Based on that, if DOM 2 doesn't suddenly change it last minute, > > I agree with an earlier comment that XSLT can apparently produce > > documents that have nothing to do with DOM, and as such, Xalan > > should be modified, not Xerces. > > > > (Of course, I'm not official here. It's just my two bits as an > > observer.) > > > > - Tom
