Richard Hamilton <[email protected]> writes: > Is the following correct?
[ Not intending to speak for Peter… ]
> - CSL encodes the details of how a particular style works. It provides
> a machine readable set of instructions that can be used by a processor
> to generate output that follows a particular citation style.
Yep.
> - To use CLS with DocBook, you could write a stylesheet that would
> take a biblioentry and format it based on the contents of a particular
> CSL file. You might do that as a pre-processor and convert biblioentry
> into bibliomixed, or you could convert directly from biblioentry into
> fo, html, etc.
Also, yep.
> - Or, at least for HTML, you could convert a biblioentry into CSL-JSON
> and convert it to HTML with citeproc-js or pandoc-citeproc. I haven’t
> tried out citeproc-js or pandoc-citeproc, so I could be way off on
> this one.
That might actually be the only short-term practical solution. I’ve
filed a few bugs on the CSL spec and its test suite. Near as I can tell,
what actually exists is a reference implementation (citeproc-js) and a
specification that only incompletely describes the behavior of that
implementation.
I still might poke at a DocBook+CSL to HTML stylesheet, but my
enthusiasm as waned significantly. A third party implementation of CSL
is going to fail tests in the test suite. The specification is not going
to reflect why those tests *should* pass, and the only recourse is going
to be to reverse engineer the citeproc-js implementation (either
literally or by making something that’s bug-compatible; in as much as I
assert that an implementation that doesn’t conform to the specification
is buggy). Kind of disappointing, really.
Be seeing you,
norm
--
Norman Tovey-Walsh <[email protected]>
https://nwalsh.com/
> The common excuse of those who bring misfortune on others is that they
> desire their good.--Vauvenargues
signature.asc
Description: PGP signature
