Re: 2.0 stylesheets and complexity

Hi,

I'm having second thoughts about this. The URIs really belong in keydefs inside of the spec. The fact that the 2.0 spec will have less than 15 of these will help (versus theÂ 60 or so of them in 1.3). I'm thinking that we could do a much simpler version of the publication metadata extraction plugin that would provide ant everything needed for URI and file naming.

Best regards,

Bob

On Thu, Nov 1, 2018 at 10:31 PM Bob Thomas <bob.thomas@tagsmiths.com> wrote:

Hi,

I've been thinking about the quagmire of complexity that I got into with the 1.3 stylesheet cover pages. My design goal was to be able to create artifacts with the correct filenames, and then to combine those outputs into a .zip archive that contained all of the artifacts in their correct places. This involved processing the information out of the DITA markup and bringing it back into ant through an xmlproperty file. This spun out of control because of the rules for determining the URIs. Paul Knight did a good job of describing how the names areÂderived various nodes in the URIs, but ultimatelyÂit became too complicated to generate reliabilityÂwithout having a deep understanding of the component pieces and when to use them.

Perhaps the xmlproperty file idea would work better in the other direction. Namely, put the URIs and filenames in an xmlpropertyÂfile so that they can be used in ant and then brought into the XSLT stylesheet as transformation parameters. The only information that would be brought in through the xmlpropertyÂfile would be URIs and filenames that vary from version to version. The rest of the cover metadata could be handled with keys.

I have attached a prototypeÂxmlpropertyÂfile and a transform parameter file to show what I have in mind.

Best regards,
Bob

--
Bob Thomas
+1 720 201 8260
Skype: bob.thomas.colorado
Instant messaging: Gmail chat (bob.thomas@tagsmiths.com) or Skype
Time zone: Mountain (GMT-7)

dita message