#barnabywaltersand I can’t find a good rust DOM implementation anywhere, which is weird seeing as servo is written in it
#barnabywaltersso it looks like writing a rust mf2 parser without relying on only questionably maintained html scraping packages will be a bit of a challenge
#[tantek]the servo parser should be solid, fairly well tested on real world content
#barnabywaltersthe parser looks fine, but the DOM implementation which comes with it is written “only for testing and security related issues will be ignored”
#barnabywaltersthere’s an issue in the repo about finding a new maintainer, and lots of people responded but it doesn’t look like anyone’s taken it over yet
#barnabywaltersI’m not so familiar with the current mf2 parsing model, is it possible to make a streaming parser which doesn’t require a full DOM in memory the whole time?
#barnabywaltersIIRC the python parser took a recursive approach, although I only worked on it a little bit right at the beginning
#barnabywaltersbut mf2 has so many special cases that it might not be possible to completely parse a document with a single tree traversal
#barnabywaltersand who knows what additional special cases have been added in the last few years, when I wasn’t paying attention ;)
#aaronpkthere haven't been too many parsing level changes
#aaronpkmost of the big changes have been in the interpretation of the parsed data, like post type discovery or authorship discovery
#barnabywaltersokay, which is out of scope for the parser anyway. good to know
#aaronpkbut now that i think about it, we're probably due for a blog post describing the actual parsing changes in the last few years
#barnabywaltersprovided the prose algorithm is up to date I can work from that, but it’d definitely be interesting to have a prose summary of what’s changed