[rfc-i] PDF production, Re: Fwd: New Version Notification for draft-flanagan-plaintext-00.txt

Julian Reschke julian.reschke at gmx.de
Sat Jun 28 04:08:28 PDT 2014


On 2014-06-27 21:43, Joe Hildebrand (jhildebr) wrote:
> It may be that a brief sketch of potential tooling might be interesting
> here.  One possible approach for generating the PDF would be to do so from
> the HTML format, using a tool that is particularly good at processing the
> CSS hints for pagination
> (http://www.w3.org/TR/CSS2/page.html#page-breaks), such as PrinceXML
> (which is why Julian pointed us to that earlier).  This approach would

<http://www.princexml.com/>

> have the benefit of keeping the look and feel ("brand") of the different
> document formats similar.  The RPC would need a copy of PrinceXML, (and
> potentially the I-D submission tool) but only people that wanted to
> generate PDFs locally would need their own copy.  Only people that wanted
> to do that without the PrinceXML watermark would need to pay for it.  One
> day when browsers implement all of the printing standards from the W3C,
> PrinceXML would no longer be needed.

True, but given the speed vendors are fixing CSS3 bugs, it might take 
time :-)

> The Table of Contents could then be generated in PDF using the
> target-counter CSS function (http://www.w3.org/TR/css3-gcpm), which
> PrinceXML supports.  This would give you exactly what you want.  Page
> numbers, and a way to index to them.

Been there, done that. See example: 
<http://greenbytes.de/tech/webdav/rfc7230.prince.pdf> (and yes, I'm 
aware of the overflowing artwork boxes).

> Note: this is just one possible approach.  I could also imagine generating
> the PDF from the XML without going through the HTML, with a good amount of
> elbow grease.

Using Apache FOP. For instance. Nothing new here: 
<http://greenbytes.de/tech/webdav/rfc7230.pdf> (note that the different 
look & feel has nothing to do with the process, so please ignore this 
for now)

> However, once you had a PDF with all of the properties you want, it might
> still be difficult for me to consume your review that only used page
> numbers as references, because I might be reading the HTML version that
> doesn't have any notion of pages.

Exactly.

Best regards, Julian



More information about the rfc-interest mailing list