Unsure how best to reach you, so opening an "info" issue.
I have refreshed the arXiv (tex-to-html) build system I maintain, and it makes a lot more articles -- loaded with TeX.js -- available for easy testing and diagnostics.
Currently the articles are unpacked from a ZIP file with all assets received from the backend, and all figures are embedded as inline data URLs. As before, TeX.js is loaded after a document.write that embeds the unpacked / styled page.
This setup seems to pose challenges with different browsers, I am wondering if it would help to serve all tex.js-related assets from the mathweb.org server?
hypothesis and mathjax both have a somewhat hiccupy behavior, certain page loads will have mathjax not start loading at all, and on Firefox the hypothesis sidebar almost never loads (and if it does, there are no active onClicks after).
Somehow there is no longer a "zoom over" feature for figures. Was it removed, or is that somehow due to me now using data urls?
I was very impressed to see the new table of contents, with bibliography items linked to http://doai.io identifiers (or if it is not new - I just discovered it! :+1: ), super cool. Tiny quibble there - the doai links could have a target='_blank' attribute set, since you never really want to navigate away from the article to see a citation.
Feel free to send me suggestions on how to improve serving the assets in that page to make TeX.js load as flawlessly as possible, I am currently investigating if I can pull the zip-to-html asset unpacking entirely in the backend, to avoid the very awkward document.write I have to resort to now.
Hi @davidar !
Unsure how best to reach you, so opening an "info" issue.
I have refreshed the arXiv (tex-to-html) build system I maintain, and it makes a lot more articles -- loaded with TeX.js -- available for easy testing and diagnostics.
It has a new URL and you can inspect the reports at: https://corpora.mathweb.org/corpus/arxmliv/tex_to_html
Individual articles now have dedicated URLs (similar to arxiv-vanity) and you can see a (random) example at: https://corpora.mathweb.org/preview/arxmliv/tex_to_html/astro-ph0001007
Several items that I am wondering about:
document.writethat embeds the unpacked / styled page.mathweb.orgserver?I was very impressed to see the new table of contents, with bibliography items linked to http://doai.io identifiers (or if it is not new - I just discovered it! :+1: ), super cool. Tiny quibble there - the doai links could have a
target='_blank'attribute set, since you never really want to navigate away from the article to see a citation.Feel free to send me suggestions on how to improve serving the assets in that page to make TeX.js load as flawlessly as possible, I am currently investigating if I can pull the zip-to-html asset unpacking entirely in the backend, to avoid the very awkward
document.writeI have to resort to now.