attachment

<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns="http://www.w3.org/TR/REC-html40"> <head> <META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=us-ascii"> <meta name=ProgId content=Word.Document> <meta name=Generator content="Microsoft Word 10"> <meta name=Originator content="Microsoft Word 10"> <link rel=File-List href="cid:filelist.xml@01C2EFA3.D170E620">  <style>  </style>  </head> <body lang=EN-US link=blue vlink=purple style='tab-interval:.5in'> <div class=Section1> <o:p> </o:p> What I am wondering is how religious we have to be about the ordering of the objects needed to image a page. Right now, PDF/is Ver. 0.5 proposes the following layout for page objects:  <o:p></o:p> <o:p> </o:p> Page Object<o:p></o:p> Content Stream, Image Mask, Image Object        %repeated as many times as necessary<o:p></o:p>                                                                         %one content stream per image or image-mask pair<o:p></o:p> Resource Dictionary<o:p></o:p> <o:p> </o:p> If the number of images or image-mask pairs on the page is not known in advance, then the Page Object cannot be definitively written until the page is completely processed since the number of content streams is not known at the time the writer encounters the Page Object. If there is only one content stream for the entire page, then the Page Object is known before any page processing has occurred, but the single content stream is not known until the page is completely processed. So place the content stream in the file after the all the images and image-mask pairs. <o:p></o:p> <o:p> </o:p> Page Object<o:p></o:p> Image Mask, Image Object         %repeated as many times as necessary<o:p></o:p> Content Stream<o:p></o:p> Resource Dictionary<o:p></o:p> <o:p> </o:p> A writer could assemble all this in memory before writing out the objects for the page-in that case, any ordering of objects is possible. The only certain constraint is that all the objects needed to image a page come together in the file. Even then there is no requirement that the Page Object be first. So the following is also plausible.<o:p></o:p> <o:p> </o:p> Content Stream<o:p></o:p> Image Mask, Image Object         %repeated as many times as necessary<o:p></o:p> Resource Dictionary<o:p></o:p> Page Object      <o:p></o:p> <o:p> </o:p> I know Ver. 0.5 has conformance rules, but I believe it is based on assumptions about the writer and reader (and the supporting libraries) that need to be made more explicit. For example, a reader could accumulate all the objects in the file and then image the page when it encountered a Page Object. Or it could be set up to image the objects in the order they are encountered, but there is no guarantee that the image objects can even be in that order so that a page buffer is required. <o:p></o:p>                                                            <o:p></o:p> Somewhere in here also is how the association is made between the content stream elements and the objects to which they apply. For example, when the content stream says "/Obj1 Do" how does it know which is Obj1? Is there a Name entry in the image object, or is it via the Resource Dictionary? <o:p></o:p> <o:p> </o:p> There has been e-mail on some of these points before, but I would like to collect all the points in a single discussion at today's meeting.<o:p></o:p> <o:p> </o:p> Rob <![if !supportLineBreakNewLine]> <![endif]><o:p></o:p> <o:p> </o:p> <o:p> </o:p> <o:p> </o:p> <o:p> </o:p> </div> </body> </html>