Issues

I can't add issues for this repository, so I'm (Iain Murray) putting some issues here, based on the experience of building the v70 ICML 2017 proceedings:

Here are bugs I've noticed in the PMLR import scripts. I may be able to help with these another time, but I don't have time to fight with ruby right now. For now, I've worked around them for these ICML proceedings.

1. { and } must not be removed from within maths $...$ sections.

2. The url field in the _posts uses the key provided in the original .bib file used for import, rather than the new key created by the import script and used everywhere else.

3. I think the id keys should be stripped of non-ascii characters.

4. Noticed later: authors that had non-ascii characters in the key didn't get links to their supplementary files added in the "Extras:" field in their _post file. Though hopefully this bug will go away if 3. is fixed.

Wishlist:

5. Ideally a bibtex_author field like I've used would be generated and used, so the bibtex entries for all volumes could be used immediately by all users.

6. Ideally there would be a documented mechanism for allowing emphasis in abstracts.

7. Document what to do with authors with "von" or "Jr" parts of their names, and perhaps deal with them differently.

8. I don't think it's legal to have < and > in html, even if inside the $...$ for mathjax. Although in practice it usually works. I replaced these with < and >, and then reverted to < and > using liquid-jekyll to generate the bibtex. (Although when the bibtex appears in pre tags I think they need changing back to < and > again?). Maybe not worth bothering with...the generated pages are far from valid html anyway...