On Thu, Jan 06, 2022 at 11:31:45AM +0100, Patrice Dumas wrote: > On Wed, Jan 05, 2022 at 09:20:28PM +0000, Gavin Smith wrote: > > > > I think mark-none would be fine. > > Ok. > > > I saw no-bullet was being used in the current output; that would be okay > > too. (See below.) > > I would prefer to replace it with more semantic classes, but we will > see. > > > But it doesn't matter that much as nothing uses these class names. > > It does, as we want to make it easy for users to develop their own CSS > styling, and for that there need to be class names, rules about classes > that make them easy to understand and use. > > > I don't know how hard it would be but I think it would be nice only > > to output the required CSS in HTML files. > > Keeping track of the CSS use is easy as all the CSS should go through > go through html_attribute_class(). Outputting the CSS block after the rest > of the file has been generated can only be done, I think, by holding off > any output before the very end, even when split. I can try to do it > that way.
I think I had a similar problem with MathJax, only referencing the JavaScript file if actually needed. > Note that a similar issue arises with LaTeX output, the packages needed > depend on the @-commands used, maybe it would be good to the something > similar for the LaTeX output to determine the header at the end. That would be a good idea as it would make the output more reliable by not relying on as many LaTeX packages.
