@JosephWright thanks for that. My last mail re font callback not quite right, the code is supposed to work if a callback already installed but apparently not, I'll try to sort it out today sometime, but I don't think it affects the release
@JosephWright the 2.8-rc3 code is using reset_callback which we didn't copy into the base code
@JosephWright I might be confused....
@JosephWright phew! confused by browser's annoying habit of renaming on download, I unpacked luaotfload.tds.zip but the thing I had just saved was luaotfload.tds(1).zip :-)
@JosephWright yes I've just re-jigged test2e to only install the modified font-callback if there is nothing currently there, which should be enough to force the same names for cm fonts as pdftex in the test suite, I've backed out the code that tries to merge that in with an existing callback. re-running full tests...
@JosephWright will mean that xetex and luatex give different font names for duplicated font loading if we ever have tests that test the new font loaders but cross that when we come to it..
@JosephWright not sure, probably it's covered by the linked question already, I didn't look that hard just waiting for the tests to run, but looking in the build directory I see I've broken all the luatex ones (or at least the font callback isn't restoring the old names) need to try again...
@DavidCarlisle Nope.... you can only blame @JosephWright for not addressing this issue right from the start when he took over the maintenance of beamer ;-)
@CarLaTeX :) I am not against the contraption itself, my fury was about people not taking care with others when using that thing! It could hurt bystanders!
@CarLaTeX ooh very interesting background! I think nowadays the vim/emacs distance is no more, as lots of plugins from both fronts make editing productive regardless of the underlying tool. I have a friend who is a vim diehard and he uses emacs with evil (extensible vi layer) support.
~
~
~ VIM - Vi IMproved
~
~ version 7.4.2367
~ by Bram Moolenaar et al.
~ Modified by <cygwin@cygwin.com>
~ Vim is open source and freely distributable
~
~ Sponsor Vim development!
~ type :help sponsor<Enter> for information
~
~ type :q<Enter> to exit
~ type :help<Enter> or <F1> for on-line help
@JosephWright in test2e we fudge fontspec.ltx so the encoding stays at OT1 but \everyjob still switches the font loader to luaotfload then the code in test2e saves that definition and wraps that with code to try to preserve font ids.....
@JosephWright well I'm not sure exactly what happens after the error: in the base almost all fonts are preloaded and basically it all seems fine but I think it's causing issues in amsmath (or at least that is where I came in) first thought was only to add the font id code if no define_font callback is installed already but that would only work if we arranged the test base format not to load luaotfload and that seems to be taking the testing further from the real format. I may just leave it a bit
@JosephWright ideally I'd allow luaotfload, preserve font ids and not break the luaotfload cache, but not actually managed that this morning:-)
I had a tex-file that basically started as follows
\documentclass[
a4paper, hyperref, amsmath, headinclude=true, ngerman, final
]{scrbook}
\usepackage{babel}
\usepackage{ntheorem}
\usepackage{graphix}
I converted this into my own custom class file
\NeedsTeXFormat{LaTeX2e}
\ProvidesClass{kit...
@DavidCarlisle My answer would be: Just do it like a normal person and don't create a fcking template, you little fcker. Sorry, Gordon Ramsay's kitchen nightmares overdose.
@JosephWright well "we" carefully arrange a config file and arrange the format is distributed to millions of users, we can't complain too much if one of those users uses it....
@DavidCarlisle Sure (though it's there mainly as KB added dvilualatex after some people complained about the lack of DVI mode even though it's clear what the LuaTeX team feel on this)
@JosephWright but patch the format or state that you need \usepackage[OT1]{fontenc} or a dvi driver that understands otf fonts as loaded by luatex. I think really doing the latter is clearer in the long run/
I just tried to install LuaTeX 1.0, but failed, because I can't compile luaotfload. »make resources« for the glyphlist fails with error 255. Has anybody succeeded in building it from source?
@DavidCarlisle LuaTeX became so much faster during the last years. I simply was curious about v. 1.0. Had no idea that there will be a new LaTeX version solving some issues with it.
@DavidCarlisle And I have no idea, neither about lollipop, nor about eplain. Let's assume I can live with it.
@KeksDose well in this case it's just working around a typo in luaotfload, but in general in the releases since 2015 we've done a lot of work making things work better in lualatex and xelatex "out of the box"
@KeksDose no one ever used lollipop, and hardly anyone except Karl ever used eplain, so you'll probably get by without them.
@DavidCarlisle -- hello fuzzy face. there are several more purple squares recently that are much better imposters than the one @egreg showed for you. (something is going weird with this general site -- the disappearing double backslashes, the random substitution of avatars, ...) not in the best shape here ... have had bad cold off and on since before christmas, and gordon now in hospital with pneumonia. wish us well, please.
@DavidCarlisle -- i'll check with karl about the ltnews. want to make sure first that we're otherwise up to date.
@yo' -- thanks, shall do so. (i do hope we weren't responsible for your cold a couple of weeks ago. we may not have shown obvious symptoms, but we were germ-laden.)
@barbarabeeton you are just using the wrong metric. It's within a couple of keys of being a correct word, so near enough correct for me. meanwhile @egreg and I agree for once: bug in amsart:-)
@DavidCarlisle -- reference for "this" bug? (plenty of known bugs in amsart. i've asked for permission to update until i'm blue in the face, but no satisfaction. i've kicked it up to the top level, jumping over several heads with no warning; let's see what happens; hoping it doesn't cost me my job.)
When I use the amsart document class and leave a subsection title empty, then the subsection number is bold. Minimal working example:
\documentclass{amsart}
\begin{document}
\section{Test}
\subsection{} Text
\end{document}
Strangely, when I include the hyperref package, subsection numbers do n...
@DavidCarlisle -- need to get editorial opinion on this. i think experience so far has been that heading text usually disappears uniformly below a certain level. but we do need to have the same result with/without hyperref, and restricted to just where there is no heading text. (the basic article model on which amsart is based is too simplistic. but we really were only trying to remain tag-compatible!
@barbarabeeton yes but see my example (posted full code in the answer) even if all subsections uniformly use empty heading text the formatting is inconsistent with the first section using medium weight number and all later ones using bold.
@barbarabeeton I added a couple of further patches. As far as I can see, line 1056 should have \gdef\@secnumfont{\bfseries} and \@seccntformat should end with code for globally resetting \@secnumfont to \mdseries.
@egreg -- i'll look at the details tomorrow, when i'm on ams time. (working on tug stuff today.) but even if valid and justifiable, i can't promise that a fix will be made any time soon, to my great distress.
@barbarabeeton i mean, for you it's not that interesting as you've got no direct flights to either Prague or something closer, but there may be people who can get to PRG directly but not to something significantly closer. And I plan to take a car...