Hi mathjax-dev,
Below are the notes for the TC meeting in September (sorry for the delay).
If you have questions as to the notes, please FIRST check out the recording on youtube for details before posting here.
The next TC meeting is scheduled for Oct 7.
Thanks,
Peter.
# 2017-09-07 MathJax TC Minutes September 2017
## Status update v3
* IN PROGRESS
* HTML output
* first large chunk landed in master
* token nodes (not mglyph)
* roots
* fractions (preliminary)
* stretchy constructions
* font data tables
* stylesheet generator
* major TODOs: sub/sup, under/over, tables
* minor TODOs: menclose, maction
* TeX parser
* refactoring is making good progress
* new approach is slowly taking over the legacy code
## Considerations for a first public alpha
* v3 code is now usable (e.g., easily packaged, usable in NodeJS or in page)
* HTML output is close enough to real life in terms of code complexity
* identify layout issues due to incompleteness which should be ignored
* code issues (errors etc) should not be ignored
* performance is now worth tracking
* decide how to publish the modules
* decide how to publish some builds for casual users
* On performance
* we can now start tracking performance in detail
* cleaner separation of mathjax and browser/layout performance
* i.e., mathjax is done and has injected output into DOM vs browser layout is actually done rendering that
* as the HTML output gained features, the slow down was more on the browser-end
* rough idea:
* 1200 equation sample (but some with missing features)
* mathjax is done in 6s
* [Addendum: ~4s but browser layout takes considerably longer]
* what kind of feedback would be useful at this stage?
* primarily: feedback on separation of parts
* e.g., is this useful for what you want to build with MathJax.
* any bugs, integration issues
* not so much: layout issues
* feedback on the general "feel" of v3 would be good, e.g.,
* the APIs as far as they already exist
* the developer workflow
* also interference, e.g., with CSS
* Is this the right time to try adapting v3 to virtual dom implementation?
* Yes!
* for writing into vdoms directly, utility/htmlnodes and utility/DOM should be the only (small) pieces you need to reimplement
* but just using e.g. react's dangerouslySetInnerHTML would be a good to try as well
* Other requests before first alpha?
* ALL: TeX input (even if still partially shimmed)
* ok if we just pull in all legacy code for the alpha
## Other development updates
* MathJax v2.7.2 is out
* please switch to it
* Mathjax-node v1.2.0 is out
* `typeset` is now a dual callback/promise API
* bug fix release ready
* minor issue with CDN URL for webfonts in stylesheet
## ACTION items
* Team: create demos
* Team: notes on HTML utility classes