@EnthusiasticEngineer That is the plain Tex format as described in the texbook it is ok for test files and very simple English language documents, not so good otherwise
@CarLaTeX did anyone tell them that they have the wrong date, that doesn't say 2019
@JosephWright wondering about adding \ifpdf to iftex useful feature or feature-creep... ?? I'd like to get ifpdf and ifluatex out of oberdiek/* (need to work out a plan about hpbsub-[generic,hyperref].sty though...
@JosephWright yes but it was first thing @UlrikeFischer asked, and ifvtex has tests for output mode, and the difference between tex and pdftex being separate engines and pdftex in dvi mode and pdftex in pdf mode being the same engine with a different switch is something that isn't clear to anyone who hasn't studied the internals. already blurred engine distinction with ifluahbtex now being true in luatex if the binary luaharfbuzz module is available. So I'm tempted but still a bit undecided..
@JosephWright yes it depends why you want the test as in "can I make it active..." I suspect that for most packages that's more useful, just as in luatex set \ifluatex true even if no luatex primitives have been enabled as that tells you you can do tex.enableprimitives and turn them on.
@JosephWright of course that's why the ifpdf package ends up with dozens of extra commands, checking the engine, which version it is and what state it is in....
@JosephWright oh I think l3build has changed, that config is similar to the one @UlrikeFischer set up in oberdiek but there as well as in iftex I now get kpathsea: Running mktexfmt pdfplain.fmt and it tries to make new formats
@UlrikeFischer Have you seen the luaotfload harfnode-dev test problem? The first harf test just freezes until Travis hits some timeout, but I'm unable to reproduce this locally. Do you have any idea what might be going on?
@UlrikeFischer I'll clean up, thanks, I wanted to get a clean slate before I try to extract ifluatex/ifvtex/ifpdf. I'm tempted to simply drop hobsub, a grep of my texlive tree shows it is not used anywhere except hyperref.
@FaheemMitha well clearly if you want to use the OpenType font you need to do that, however there are tools to extract 256 character subsets to type1+tfm for pdf
@FaheemMitha no apart from the fact that it can not read the font format opentype fonts often have thousands of characters and pdftex fonts can have at most 256
@FaheemMitha well not really. I just made a calibri adaption for pdflatex with autoinst for someone. But you have to make the adaption, if you want something "that simply works" you need such an engine.
@DavidCarlisle I don't understand your attachfile2 failure. The code there uses \the\year \the\month etc. If I compile directly I get 20191026195600, but with l3build 20160520106000. So it looks as if l3build is setting the date. Why do you get another one?
@DavidCarlisle no idea why line breaking changed in ifluatex, but it looks harmless. The kvoptions is clearly an oversight. The askinclude-tests have a clear failure - this was already there a few weeks ago - and I have not idea what is breaking and why as this test macros are so curious.
@JosephWright then why don't I see the change when I use the github version? Which function does change year etc?
@FaheemMitha -- On the old hot-metal Linotype machine, the upright and italic alphabets for a family were mechanically in parallel, so a letter in italic would have the same width as that same letter in the upright form. That could be considered "metric-compatible", but the appearance would surely be different.
@FaheemMitha lots of fonts are commissioned by Microsoft
@UlrikeFischer well I haven't prepared that PL as I was working on iftex then was going to do oberdiek to extract ifluatex, but I may work on the PL2 code first/
@MarcelKrüger I don't really now what happens in the config-harf tests, I tried to exclude the failing tests but then the next one don't work. I can't continue now, but will try tomorrow afternoon,
@FaheemMitha -- Actually, some Microsoft fonts are rather nice. They did a good job with Cambria (their math font), but of course it looks best when used with a compatible text font, and whether or not one likes the effect is subjective. (I'm not a Microsoft devotee, in fact, rather the opposite, but they do deserve credit for some things.)
@DavidCarlisle -- The OpenType Math specification is largely based on Knuth's Appendix G, with one significant omission, the parameter on which amsmath's \genfrac depends. This causes problems dealing with sizing of fences (among a few other things), but that has been mostly handled now by workarounds.
@DavidCarlisle -- There's actually a bad design feature that shows up in \genfrac -- the thickness of the fraction rule depends on the same font parameter as the clearance of the numerator and denominator from the fraction rule. Not wanted, but ultimately the result of limited memory for assigning font parameters in a .tfm file. Too bad. A genuine "feature" now.
@DavidCarlisle -- The OpenType math table misses exactly one of the tfm math parameters, and that's the one. I forget which parameter it is, but researched it carefully at the time, and checked with Murray Sargent; he said "they didn't need it", so left it out.
@barbarabeeton yes they don't have the parameters for the special genfrac delim but (if I recall correctly) they have separate parameters for spacing and rule thickness in fractions
@DavidCarlisle -- Those may be there now, but I'm pretty sure they weren't when I checked. That was several years ago, and I don't have access to the correspondence. (Although I hope to regain that sometime.)
@FaheemMitha -- I think it's documented somewhere in a publicly available Microsoft document that they visited Knuth to discuss the mechanics used by TeX, which is essentially Appendix G.
