datatracker icon indicating copy to clipboard operation
datatracker copied to clipboard

Published 20 hours ago • verified publisher icon ietf-tools

Draft submission is stripping SVG from PDF.

Open woobagooba opened this issue 3 years ago • 3 comments
trafficstars

After discovering that direct kramdown submission is not enabled, a conversion was made (using author tools) to XML and then the XML was submitted (unaltered).

https://datatracker.ietf.org/doc/draft-duffy-csmp/

The PDF is stripping all SVG graphics.

The PDF suspiciously looks identical to the text rendering.

Author tools rendering (from the kramdown) of the SVG within the PDF is fine.

There appears to be a problem either with the author tools conversion from kramdown to XML, or DataTracker's rendering of the PDF.

woobagooba avatar Aug 17 '22 18:08 woobagooba

Datatracker generates PDFs from htmlized output. These are different from the PDFs generated by xml2rfc.

kesara avatar Aug 17 '22 21:08 kesara

There are several things that you need to know to understand what has happened here.

  1. kramdown-rfc, right now, does not produce v3 rfcxml - it produces v2-ish rfcxml and relies on going through the v2v3 conversion. Carsten may change this as we improve various things (v3 grammar issues, and how easy it is to get the bibxml content).

  2. The pdf that you are looking at is exactly as you guess - it is a pdf rendering of the htmlization (not that I didn't say html rendering) of the text document. The datatracker labels this with a pdf button right now - it should likely label it with a button that says 'pdfized' with a tools tip of 'pdf of htmlized text' or the like. We've already done this for `htmlized'.

  3. If you had actual full v3 source, you would be able to (and I think the datatracker does for you) render v3 html and pdf output as defined by the 799x documents. This is what you are seeing from authortools when you ask it for the pdf, since it is running --v2v3 and then xml2rfc --pdf.

We could look at a policy change at the datatracker where we render the v3 outputs for v2 input text (which we already run through --v2v3 anyhow before rendering text), and then we would get 'pdfized' buttons for pdf htmlized text, a pdf button that leads to v3 era pdf.

cc @larseggert

Older drafts/rfcs would still only get what we can (htmlized/pdfized) build when we only have the text available as an input.

rjsparks avatar Aug 17 '22 21:08 rjsparks

Thank you Robert. Yes please to align author tools and the submission process. Author tools PDF is the only format properly rending my SVG.

woobagooba avatar Aug 18 '22 00:08 woobagooba