🚀 go-pugleaf

I went to Canada for a few days, so I haven't had time to finish looking into this. But I guess I should post an update to let people know what I've found so far. 

Examining the files Alyse sent me shows that there is a named destination of the form G###### (where # is a digit) with no assigned page.. If I assign a page to that destination, clicking the broken link causes a jump to that page. 

Redistilling the source file results in an undefined destination error of the same name in the Distiller progress window and log file, indicating that the intermediate postscript file included a pdfMark command to create the link but no corresponding command to create its destination. 

Because I already knew that G###### destinations correspond to Frame paragraphs (where ###### is the Frame ID for the paragraph object), I concluded that a Frame cross-reference is converted to a PDF link to a named destination for the paragraph that contained the cross-reference marker. 

Examining the intermediate postscript files shows that they contain two kinds of statements that are potentially synonyms for the pdfMark statements for G###### destinations. The first kind end with FmND and always produce named destinations. The second kind end with FmPA and only produce destinations if the  Frame document's Create Destinations for all Paragraphs PDF Setup option is enabled. 

If that option is unchecked, creating a cross-reference to a paragraph and distilling the file results in an FmND postscript statement. Copying and pasting them to another Frame file and distilling usually results in an FmPA statement, which yields a broken link (undefined destination). Changing the FmPA statement to FmND fixes the problem, but isn't a workable solution. 

So the question is, under what circumstances does Frame convert the marker to a FmND versus an FmPA statement?  

Saving the file that produces a functioning link to MIF creates a pdf section for the target paragraph that includes a <pgfReferenced 'yes'> statement. Saving the file that produces a broken link to MIF produces a file that lacks that statement. I supposed that the FmPA statements are generated for paragraphs that are missing that property, but it turns out they always do when you cut and paste, even though cutting and pasting sometimes gives a functioning link. So apparently the only way to determine the circumstances under which cutting and pasting cross-references will create working PDF links or not is to try different combinations and distill the resulting files. 

I haven't had time to do that  yet. However, because it depends only on whether or not Frame generates an FmPA or FmND postscript command, I believe we can conclude that the version of Distiller you use has nothing to do with it.