@JosephWright I have the suspect that someone is serially upvoting me once a day, what should I do?

@CarLaTeX Just wait 24 h and see if it gets reversed

@JosephWright Is there going to be an update for l3backend soon?

@HenriMenke Yes: I'll sort today

@JosephWright Thank you!

quack

@DavidCarlisle You should run lualatex with the -safer option, because the following currently works on LaTeXCGI

% !TEX lualatex
\documentclass{article}
\begin{document}
\directlua{
local http = require("socket.http")
local ltn12 = require("ltn12")
local t = {}
http.request{
    url = "http://www.cs.princeton.edu/\string~diego/professional/luasocket/http.html",
    sink = ltn12.sink.table(t)
}
tex.sprint(-2,table.concat(t))
}
\end{document}

@DavidCarlisle The barrier is maybe a bit higher but you don't want people to abuse LaTeXCGI for command and control of their botnets.

@HenriMenke ooh

@DavidCarlisle Or maybe the -nosocket option is sufficient. The -safer will also disable shell escape. Restricted shell escape should be fine. You just don't want people to run curl.

@HenriMenke you can't use --safer, then fonts no longer work. Does --nosocket avoid this?

- If parsing global options only remove from `\@unusedoptionlist` and ignore unknown keys
- If parsing local (`opt@\@currname.\@currext`) options:
    - If in cls:
        - If `\@classoptionslist` is not `\relax` and equal to `opt@\@currname.\@currext` add unknown keys to `\@unusedoptionlist`
        - Else ignore unknown keys
    - If in pkg throw errors on unknown keys

@HenriMenke I am currently working on PGF's dvips driver to make it compatible with Ghostscripts new transparency operators (as of version 9.52). Going to make PR soon.

@JosephWright does that look like reasonable parsing rules for unknown keys^^^

@HenriMenke yes... or probably if we want finer control I could probably adjust the lua path in loading in the format to prevent those modules being loaded? but nosocket sounds like a good first move

@Skillmonlikestopanswers.xyz Seems OK

@Skillmonlikestopanswers.xyz Oh sorry for sneaking in between.

@AlexG not a problem :)

@DavidCarlisle @UlrikeFischer You should have received mail.

@AlexG Oooh

@JosephWright ooh

@AlexG and you sneaked in with some very interesting news :)

@Skillmonlikestopanswers.xyz ooh

@JosephWright Also implementing Blend modes and Transparency groups.

@AlexG Sounds good

@JosephWright also it is possible to parse the \@classoptionslist within a class with the rule set of "parsing global options" (since a class might be loaded using \LoadClass and get its options via \PassOptions... and hence might have local options differing from global ones).

(The only thing I'm not yet lucky with is my loop to remove from the \@unusedoptionlist, which currently only removes the first matching option, I should perhaps change that to removing every match)

(also active commas aren't supported, but that shouldn't be an issue, people activating commas before \begin{document} should be punished)

@DavidCarlisle looks good. It seems to overwrite my settings for this page, but it does indeed work out :)

@DavidCarlisle ?

@yo' I guess it's about the new setup for running LaTeX on learn-latex, which is somewhat comparable to Overleaf.

@Skillmonlikestopanswers.xyz ah

@yo' well, minus the collaborative stuff :)

@Skillmonlikestopanswers.xyz yeah, and minus a lot of other things :-)

@yo' insignificant!

@yo' but with the plus that every error is an error, you can't ignore them ;-)

@UlrikeFischer that's a toughie really.

@UlrikeFischer and you really see the logs if there is an error!

ooh errors

@Skillmonlikestopanswers.xyz yes, that is very nice. I used that yesterday to check what luaotfload it doing, a \blub here and there was very useful ;-).

For the first time, I didn't installed TL in /usr/local!

@DavidCarlisle twitter.com/rosielanners/status/1277545491052584960

@TooFatManNoNeck That is somewhat quick/insane. But he forgot to TeX his answers.

% \HiraganaDecList
\newcommand{\HiraganaDecList}{}% Name reservieren
\let\HiraganaDecList=\empty% Liste erstellen
\makeatletter
\foreach \n  in {0,...,30}{%needed: 0,...,47
\pgfplotstablegetelem{\n}{Dec}\of\HiraganaData%
  \ifx\empty\HiraganaDecList{} \protected@xdef\HiraganaDecList{\pgfplotsretval}%
  \else \protected@xdef\HiraganaDecList{\HiraganaDecList,\pgfplotsretval}%
\fi
}\makeatother
Show: \HiraganaDecList

@cis well as you keep it secret what is in HiraganaData, ...

@UlrikeFischer, it is working indeed, thanks for the info, though I am surprised that so much # are required ...

@BambOo well basically you are in an argument of a command which is in the argument of a command, in such places getting the numbers of # right can get quite challenging. I would suggest not to use sci generic directly but to use a style instead.

@UlrikeFischer, your recommendation is duely noted

@JosephWright ok thks

:54794838

Dec,     Hex,      Strokes,   Symb,   Descr
12354,   U+3042,   3,         あ,      HIRAGANA LETTER A
12356,   U+3044,   2,         い,      HIRAGANA LETTER I
12358,   U+3046,   2,         う,      HIRAGANA LETTER U
12360,   U+3048,   2,         え,      HIRAGANA LETTER E
12362,   U+304A,   3,         お,      HIRAGANA LETTER O
12363,   U+304B,   3,         か,      HIRAGANA LETTER KA
12365,   U+304D,   3,         き,      HIRAGANA LETTER KI
12367,   U+304F,   1,         く,      HIRAGANA LETTER KU
12369,   U+3051,   3,         け,      HIRAGANA LETTER KE

@UlrikeFischer

it's about a list of the kind

@UlrikeFischer

@cis sorry but I don't have the time to disentangle your code. Ask a proper question on the main site.

@yo' suggestion for overleaf: if there are compilation errors, perhaps change the background color or similar. There seems to be a lot of users who either don't know what the red zone with the numbers mean, og just ignores it as they are getting a PDF anyways.

@daleif ... and then ask questions like "my document compiles fine on overleaf, but if I download it throws errors" or "my document compiles, but the layout is strange". The hardest part is not to solve the problems, but to convince these users that their documents did not compile on overleaf.

@samcarter_is_at_topanswers.xyz You both have a point and I'll make sure we discuss this again internally. I'm in a meeting now, but later I'll share some history of this with you to get the context; it looks easy when you're a TeXpert, but it's not easy when you have 6M users.

@yo' I'm sure it's not easy and if you would not show the pdf, your support would surely be flooded, but leaving the users unaware that their pdf is possible malformed, is also not that good. Personally I think a pop-up would be helpful, something along the lines of "There have been errors during the compilation. Show log file - Proceed to the pdf anyway". This way the users could at least be blamed for ignoring error messages.

@yo' we know, but it should be really more difficult for the users to ignore them. E.g. some dialog they have to click away. Or some setting in the profile behind some security questions ("Do you REALLY want to ignore perhaps fatal errors???").

@UlrikeFischer I would extend this message "... ignore perhaps fatal errors. Each time you ignore a latex error, a small duckling starts to cry"

@samcarter_is_at_topanswers.xyz ;-). Which reminds me, did you see who will take part at tug2020?

@samcarter_is_at_topanswers.xyz exactly

@UlrikeFischer this idea is being pushed by me actually (and some others on the TeXpert side of Overleaf...)

@UlrikeFischer Yessss, only one participant with a photo attached. Has to be the VIP :)

@samcarter_is_at_topanswers.xyz there is second VIP later on ...

@UlrikeFischer let me check again

@UlrikeFischer Oh, the professor will also be there!

@yo' I'll keep my fingers crossed that you succeed!

@samcarter_is_at_topanswers.xyz :-)

@yo' Such a dialogue could even link to some site with debugging tips, where to get help etc.

@samcarter_is_at_topanswers.xyz Yeah. Well, we even have a standard reply we use on Support when we see a project with a lot of errors. (I think the record holder is a project which timed out with 5j errors. When a colleague increased the compilation time for the project, it compiled "successfully" with 8000 errors.)

@samcarter_is_at_topanswers.xyz ooh

Btw, for the history of this feature: Overleaf-v1/WriteLaTeX used to be on strict, whereas ShareLaTeX was on nonstop. When the two merged, it was -- from what I understood -- a tough decision what to keep. Because on one hand, strict is more reasonable (as we all agree!), but on the other hand, you don't want ShareLaTeX users left with projects that "worked" (from their point of view) and don't work anymore. And last but not least, you don't want to have projects at either option randomly

@PauloCereda There are clearly some advantages of moving TUG2020 online. Who knows if someone as busy as the professor would otherwise have been able to attend.

(or seemingly randomly, once they forget that a specific project was from ShareLaTeX).

@samcarter_is_at_topanswers.xyz one could lure him by using chocolate :)

One thing I'd like to see is a per-project option for strict/nonstop, with strict being the default for new projects.

@yo' Oh, one for the wall of fame :)

@samcarter_is_at_topanswers.xyz yeah, I think it's a record-breaker.

@yo' It would have been terrible for the acceptance of the users if the change from sharelatex to overleaf would suddenly have changed the behaviour.

@PauloCereda Maybe first check with @barbarabeeton if chocolate is safe for ducks :)

@yo' now I'm a bit confused. What is strict mode in this context?

@samcarter_is_at_topanswers.xyz yeah, exactly. You need backwards compatibility for things like this. That's also a reason why Overleaf keeps 6 TeX Lives available.

@daleif strict = stop after first error and only show the first error. nonstop = continue running until the compilation terminates, and show all errors.

@yo' btw: know that the 6 texlives are available to everyone this is great for testing!

And there are more things to consider: Should the first compilation be nonstop(so that it more easily deals with corrupted auxs) and the subsequence ones be strict?

@yo' ahh, -halt-on-error, as a heavy latexmk -pvc user, I switch to that one some time ago. Much better

@samcarter_is_at_topanswers.xyz interesting use case :)

@daleif yeah, I use the terminology from \strictmode and \nonstopmode primitives.

@yo' do you inject those in the compilation?

or use the cli options

@daleif we use the options to latexmk

@samcarter_is_at_topanswers.xyz -- I really don't knoew whether chocolate is safe for ducks, It is certainly not safe for dogs. (But it is safe for me!)

btw, I have a latemkrc file on my desktop with just three lines, that I pull in projects when I do testing and it seems to be error-heavy:

$pdflatex = 'pdflatex --halt-on-error %O %S';
$xelatex = 'xelatex --halt-on-error %O %S';
$lualatex = 'lualatex --halt-on-error %O %S';

@yo' It came really handy when a beamer user recently claimed: "this used to work a week ago" and it turned out that even TL2014 gives the same result as today

@barbarabeeton ... but are you safe for the chocolate ??? I'm certainly not

BRB we had a hiccup in Overleaf,

@yo' It is going to be interesting when Joseph finished the next siunitx. Had user on Overleaf who has a lot of issues with his thesis so I helped him out. He insisted that he needed 80+ tables in his appendix. siunitx caused them to time out on Overleaf. The alpha version of siunix with everything switch on was at least 4 times as fast.

@samcarter_is_at_topanswers.xyz -- Probably not. If it looks tasty, I will probably try to devour it.

@samcarter_is_at_topanswers.xyz Just had a question on the site, where I just check the MWE using TL15-20, and they all gave the same answer.

@barbarabeeton :) TUG'20 needs a virtual conference dinner with chocolate tasting

@yo' -- I realize I'm archaic, but am I really the only person around who still runs (La)TeX from the command line, and corrects typos in control sequences interactively?

@barbarabeeton most likely. I used to do that

@samcarter_is_at_topanswers.xyz -- I like that idea! There's a local company that creates luscious chocolate-robed ginger. Yum!

@yo' -- But it's so efficient at avoiding "nonsense" errors caused by getting out of synch.

@barbarabeeton Sounds yummy! I once got a "Travel around the world" chocolate set. It consists of small bars made from beans from different countries and they all tasted really different. The one from Brazil was really good! (cc @PauloCereda)

@barbarabeeton The blessing and curse of all these IDE which don't allow direct interaction with the command line

@barbarabeeton well, that's one of the reasons why not to use strictmode. The thing is, if you use the interactive mode, you have to do the correction twice. That's a huge annoyance to me, and also it doesn't save much time lately with computers being faster and faster.

@samcarter_is_at_topanswers.xyz KInda like if you know someone going to Japan, have them send to samples of the Japanese Kitkats, it is nuts so many flavors they have.

@daleif Oh, a friend of mine brought green kitkats from Japan - they looked funny :)

(I think they were tea-flavoured or something)

@yo' -- I guess you don't tend to make errors that turn everything that follows into nonsense.

@barbarabeeton well, I do. But more often I write \alpha as \alpah or \cdot y as \cdoty

@yo' -- Sadly, my misspellings often occur in environment names. (But, of course, they're not generally correctable interactively.)

@barbarabeeton ah that's too bad :( I try to double-check these

@samcarter_is_at_topanswers.xyz ooh

@samcarter_is_at_topanswers.xyz probably matcha, that is one classic. Some canadian people I follow online (they do various things on YouTube and Twitch) make a video from time to time opening fan mail. A fan was in Japan and had asked if they wanted something. One of the replied Send us some funny kitkats, but forget about the normal ones get the really dank sh** I think the person sent them like 20+ different flavors. I can find the video of them opening them

@barbarabeeton nope, depending on what I'm doing I correct them interactively as well (but most of the time I run in -halt-on-error)

@yo' I remember downloading the projects and compiling them on my machine as a first step for support requests...

@Skillmonlikestopanswers.xyz -- How do you interactively correct an environment when the error message comes up with an @-laden \cs?

@daleif That's sounds like a funny thing to watch :)

@Skillmonlikestopanswers.xyz I don't do that unless competely necessary

@barbarabeeton I never do, when I run in interactive mode I'm debugging low level code most of the time in which case there are no misspelled environments, as there are no environments... :)

@yo' well, many of the queries I tackled made it necessary, of course for really small things I didn't do that. But my workflow is so much modelled on my VIM, that I find debugging in Overleaf really hard.

Could sb. say me why this document crashes?

This is LuaHBTeX, Version 1.12.0 (TeX Live 2020/W32TeX)  (format=lualatex 2020.4.11)  29 JUN 2020 18:05
 restricted system commands enabled.
**test2.tex
(./test2.tex
LaTeX2e <2020-02-02> patch level 5
Lua module: luaotfload-main 2020-02-02 3.12 luaotfload entry point
Lua module: luaotfload-init 2020-02-02 3.12 luaotfload submodule / initializatio
n
Lua module: lualibs 2020-02-02 2.70 ConTeXt Lua standard libraries.
Lua module: lualibs-extended 2020-02-02 2.70 ConTeXt Lua libraries -- extended c

@samcarter_is_at_topanswers.xyz youtu.be/C0h9sII_G3A from 47:30, I think there one more but did not bother look it up.

@cis ...er... I do not think that posting 1000+ line here in the chat can help anybody help you. I suggest you try an MWE on the main site (you can use a gist if the file is really long). Without being abe to reproduce the problem, it is almost impossible to guess it...

@JosephWright github.com/pgf-tikz/pgf/pull/890

@cis I rapidly scrolled through it and I could not find nay error. On the other hand, in such a long log, I could have missed it.

@Rmano OK, thx.

@Rmano -- Scrolling isn't the best way to examine a log. Bring it up in your editor and search for "error" (or "Error" if you can't do a case-insensitive search). Then search for "warning". If you don't find either, then something is really wrong.

@cis the log stops in the middle, so probably you killed the compilation before it could write the interesting stuff in it. But beside this I quite agree with @Rmano.

@JosephWright the link commands, e.g. \__pdf_backend_link_begin_user:nnw errors for some backends (pdfmode + dvips) when used in vertical mode. So I think somewhere there should be a \leavevmode. But I'm unsure if it should go in the backend or in the user commands. What do you think?

@barbarabeeton Yes, I agree --- I normally do a copy and I have a small macro in my editor to open it in a new buffer. But my point was that the chat it's not the best place for a long test. Pasting it in a gist or in a temporary online clipboard and posting the link it's much better...

We are back to strict lockdown... :(

@PauloCereda :-( here it seems it's going better. No new cases in Northern Spain today, several localized outbreaks that seem well controlled. Finger crossed.

@JosephWright is the limitation that \clist_remove_all:Nn might fail for token lists containing braces still valid?

@PauloCereda BTW you probably missed this...

@Rmano oh sorry, I didn't get the notification! Please please register! I believe all keynotes will be recorded and be available later on.

@PauloCereda done!

@Rmano yay

@PauloCereda -- Rhode Island is one of two states in the U.S. with a continuously dropping rate of infection and hospitalization, and some restrictions are being loosened, but face masks are still mandatory, and numbers of people in one place are still limited. The governor is just now giving her thrice-weekly info/pep talk -- "you've done a good job; don't let your guard down now." Keep wearing those masks.

@barbarabeeton good to hear!

@PauloCereda and that's bad to hear!

Have a look on the Stroke Order Starpoint Map preview:

But more is coming.

@Skillmonlikestopanswers.xyz

Zoom to see the red point.

@cis looking good :)

@Skillmonlikestopanswers.xyz it doesn't override your settings for application/pdf, it just doesn't serve any PDF, only javascript (PDF.js?file=document.pdf

@DavidCarlisle in that case PDF.js is really good at faking the looks of FF's internal PDF viewer... (or does FF use PDF.js internally?)

@Skillmonlikestopanswers.xyz it is firefox's internal pdf viewer

@DavidCarlisle ahhhh :)

@DavidCarlisle in that case: never mind :) Looking good on my setup.

@Skillmonlikestopanswers.xyz it was originally a third party library then firefox adopted it but it is available from their site still as a separate download

@Skillmonlikestopanswers.xyz thanks for confirming it works. Not really quite ready for a public release yet a few loose ends to tidy up but having pdf.js means it works on mobile (or for you and others using external pdf readers) and its got biber working unlike the one we are currently using only problem is that the host will vanish in a year:-)

@DavidCarlisle why will the host vanish?!

@Skillmonlikestopanswers.xyz because amazon give you an ec2 instance for a year then want money

@Symbol1 mathptmx is ancient and does not have bold. use newtxmath or stix2 if you want times-like math.

@DavidCarlisle how greedy of them :) Well, that's future us' problem :)

@DavidCarlisle there are noticeable differences between the Greek letters of newtxmath and Times New Roman maths symbols, iirc. (that's why I use mathptmx despite its flaws)

@Skillmonlikestopanswers.xyz well don't use bold then:-)

Springer assigns that package, saying that line/page-breaking will be closer to the final output.

@DavidCarlisle I know (and it bothers me more than you could ever imagine). I did patch pmb in that document to stack 14 symbols instead of 3 to give a slightly better result (still unbearable) just for emergency cases.

I don't use bold math at all, so I guess I'm fine.

It is just that the warning messages are annoying (since I resolved all other warnings).

@Symbol1 you didn't show the warning but presumably it was saying OT1/ptm/bx/n substitued by OT1/ptm/m/n which just means you get non-bold letters for bold, but since springer will not use those fonts anyway it doesn't really matter

Friends, sorry for the language, but this is great: twitter.com/vartian/status/1277654478196551680

(There is no markdown in chat... why am I trying to format things...)

@Symbol1 There is, it just don't work with multiline... :(

Also I fail English

:D

@Symbol1 that is normal. Don't worry simply ignore the info.

Cool. I'll just ignore. (Plus it might get rejected and I'll send it to other publishers.)

Thank you all =)

@PauloCereda twitter.com/JMR_Green_Ducks