agitter
commented
7 years ago I tested the example GFM table from this guide and it worked in the HTML and PDF outputs:
You can try it yourself before making a pull request by creating the conda environment and running the build script. It's easy to set up if you have Anaconda installed.
The bigger question for @dhimmel (and maybe me) is how we want to handle table numbering, and more broadly how general the build process should be. It should be possible to automate with Jinja2. I imagine that in the markdown we could write Table {{tables.network-architectures}} and have it be automatically numbered as Table 1 in the output. We should move this discussion to another issue if we want to pursue it, but briefly we could have references.py call a function similar to get_references_from_text to get all table references. Then we could number those references by their first appearance. What gets a little more complicated is if we want to treat table references in the text differently from the actual tables, which would help keep the numbering stable regardless of table placement. We'll need the same thing for figures if we add those (#354).
@yfpeng I suggest that you initially hard code the table number.
dhimmel
I am writing a subsection of deep learning for biomedical text mining. I need to add a table but I am not sure if our script supports it because tables aren't part of the core Markdown spec, but they are part of GFM.