-t epub3 from markdown produces "invalid ids" as reported by calibre

[georgernstgraf]

Greetings folks,

These invalid ids are created from many headings i have in my documents. Calibre
offers a "fix automatically" button which then replaces all those ids with
randomly generated ones, and then the check yields no errors anymore.

As asked, I am not filing an issue yet, but I think this here is worth one.

From my Makefile:

ebook.epub::
	pandoc --metadata-file=metadata.yaml --toc --toc-depth=1 -t epub3 -i [0-9][0-9][0-9]*md -o $@

See my screenshot:

[jgm]

As I understand it, these are not invalid IDs. epubcheck will not flag them as such.
The epub spec calls for XML-serialized HTML5, and characters like ü are allowed in HTML5 identifiers. It may be that there are readers that have trouble with them, but that is different from claiming that they are "invalid."

If this is a serious problem, though, we could consider adjusting our ids to be ASCII only. Comments welcome.

[georgernstgraf]

Thank you! Importing the generated ebook into my tolino shine https://mytolino.com/products/tolino-shine-4/ (I have a slightly older model but with very recent firmware) also generates a warning, but I am able to open and read the content.

So no, not serious.

[jgm]

Are you generating the epub from pandoc's markdown? If so, try with the ascii_identifiers extension: pandoc -f markdown+ascii_identifiers -t epub3.
I think this will solve the problem.

[georgernstgraf]

Yes it does .. sorry I missed that. Awesome work!