@UlrikeFischer If you look back, I had exactly the same with v1 and v2 - lots and lots of bug fix releases
@UlrikeFischer v3.0.15 just went to CTAN
@UlrikeFischer I'm pondering how much to put in v3.1.0 - the complex number upgrade looks nice and easy ...
@UlrikeFischer I'm mainly happy so far that the change of how I do fonts hasn't really been an issue - it's all been about oversights or subtle bugs, almost all in the number code
I was looking for a printed edition of a LaTeX book in the university library and found this manual bit.ly/3gMF9yJ, @DavidCarlisle do you recommend it? :P
@JosephWright I'm afraid, I think I've explained myself incorrectly: what I found strange was the fact that the digital version of this book on amazon.co.uk costs £54, while the same version on amazon.it costs €38. Quite a difference for being the same book
@UlrikeFischer Looking at tex.stackexchange.com/questions/602271/… I'm thinking about merging luaotfload's score-dev branch to fix all these issues. I'm a bit worried though that it breaks stuff. Does it work on your system?
@UlrikeFischer It rewrites the classification logic for regular/bold/italic/bolditalic fonts to ensure that: Every family has a regular font and bold/italic fonts are bold/italic relative to the regular font.
Therefore it allows e.g. to load Latin Modern Roman Slanted by fontname using fontspec since it no longer complains that no font in the family is regular.
But since it completely changes the detection code, there is the risk of it breaking while analyzing weird fonts (e.g. I just fixed an issue which occured only for a font with italic angle higher than 17000) and it might select different font variants by default, thereby changing documents.
@JosephWright the "issue" is that \keyval_parse:NNn has the rule that there can only be one equals sign on top level per pair (expkv treats that differently, assuming that the first equals sign is the separator), and though [] adds a pair of brackets (similar enough like braces when it comes to optional arguments) for the user's eye that isn't regarded as sufficient protection for \keyval_parse:NNn. So he suggests using expkv's parser if one wants to nest key=value macros.
@JosephWright either that, or his own package strictex, that will replace every [ with [{ and every ] with }] with LuaTeX's input buffer manipulations.
@JosephWright oh, and: aaww for the part in parentheses :)
@DavidCarlisle I did, the only person who matters in this regard to me :D