@PauloCereda biblegateway.com/passage/… maybe?

Might be of interest to anyone here doing computational work:

@StefanKottwitz I did say it in the past, and I'll say it again: If the people in this chat feel my user name is bugging them, I'll change it back. As long as its just a single voice I don't really see the point though.

@JosephWright there is an error in the backend of dvips and (x)dvipdfmx: It uses __kernel_backend_select:n instead of __color_backend_select:n:

\cs_new_protected:Npn \__color_backend_family_separation:nn #1#2
  { \__kernel_backend_select:n {#1} }

@JosephWright there is also something wrong the actual implementation for dvips, ps2pdf complains about the l3pdfcol1 (xelatex can't work anyway until the binary is updated). But I can only look in the afternoon.

@UlrikeFischer Fixed

@JosephWright yes saw it. I will try to figure out the separation part. There is also the "named" model in color.sty, and the commands in dvipsnam.def. They setup names, but not spotcolors, but cmyk names. I'm not sure how this should be handled.

@UlrikeFischer My feeling is to handle that at the macro layer: we can just collect up the names

@UlrikeFischer You'll try to work out why ps2pdf is not happy?

@UlrikeFischer the intention was that those names were spot colours (or at least driver defined ) in drivers that could support them with cmyk fallback if not

@DavidPurton ooh ostriches

@DavidCarlisle well I checked what happens with them when converting to pdf, and there you get simply the cmyk. So there are at most spotcolors directly in the postscript (but I don't know if any postscript RIP would take them as such).

@UlrikeFischer yes for pdf at the time I didn't know what else to do with them. For dvips the default postscript header file converted them to cmyk but the intention was that you could use custom header files to separate out the colours. but apart from commercial print shops using textures I doubt anyone did

@JosephWright collecting the values is not the problem. The color is written differently to the postscript. So one probably need a different backend command here.

@JosephWright yes, I will try.

quack

Friends, it's so challenging to prepare a talk on something you never publicly talked about!

The vegan mafia only warns once

@PauloCereda you don't need to worry about anything after Sunday dinner

@DavidCarlisle oh no

If you look at travis-ci.org/account/repositories can you see you "organisation" repos, I can only see my personal ones not latex,3, not work :(

@DavidCarlisle oh I need to login

@PauloCereda I need access to the latex travis settings so I can enable this for @egreg

@DavidCarlisle LOL

@PauloCereda actually I need to get in for work but it seems to have changed a bit I wonder if they are forcing new repos to travis-ci.com now not .org

@DavidCarlisle for work, really?!

@DavidCarlisle that's odd, usually the org thingy was for opensource stuff and com for paid thingies.

@PauloCereda they are moving everything to .com (slowly)

@PauloCereda github.com/orgs/numericalalgorithmsgroup

@DavidCarlisle ooh nag

@DavidCarlisle can you add me to the org too?

:)

@DavidCarlisle oh no, you are not in NAG

@PauloCereda can you see your gh organisations in travis? It says I am not in any, but following the link just takes me to the github page that says I am in lots

@DavidCarlisle let me check

@DavidCarlisle ^^ /all raise, UK-TUG

We need a UK-TUG anthem

hmm I think I'll try to log out and re-connect

@PauloCereda s/anthem/requiem/

@DavidCarlisle nah, how about youtube.com/watch?v=m4zyfwtDIHE

@DavidCarlisle ooh

@PauloCereda fixed ^

The English have a user group and you can apply / despite the bloody group is about to die

@DavidCarlisle so reboot your machine is always the right way of doing things :)

@PauloCereda more or less, just logged out of travis and logged in again

@DavidCarlisle ooh let's blame someone

@PauloCereda @UlrikeFischer

@DavidCarlisle oh no, poor Ulrike

@DavidCarlisle Oooh

@PauloCereda, @DavidCarlisle Where do you see that?

@JosephWright it's a secret :) nah, it's here: travis-ci.org/account/repositories

@JosephWright they are the organisations that travis knows I am in, oh the other duck just gave a link

@DavidCarlisle <3

@UlrikeFischer I think CIELAB is now working :)

@UlrikeFischer Next is DeviceN I guess

@UlrikeFischer But first, CTAN

@JosephWright I'm pondering about postscript and dvips.

@UlrikeFischer Cool

@UlrikeFischer We have no guide from colorspace, so it's going to be trickier than PDF mode

@UlrikeFischer, @DavidCarlisle, @PhelypeOleinik, @MarcelKrüger I'll do the expl3 release before lunch

@JosephWright yes ;-(

My PC is being really pushed by all this remote working stuff, graphic tablet, etc.: need to convince someone to buy me the latest XPS

:)

@JosephWright uea?

@JosephWright or uktug :-)

@DavidCarlisle Not likely

@DavidCarlisle See Skype

@DavidCarlisle Nice idea, you can imagine the reaction (there, a Mac might actually make sense for cross-platform work)

@DavidCarlisle Don't worry, XPS 9630 is pretty good: i5 and 8Gb RAM is actually quite workable :)

@Skillmonlikestopanswers.xyz I know chat people are friendly, tolerant, may let you do whatever floats your boat, so you won't be bothered about topanswers-spam

@Skillmonlikestopanswers.xyz When we talked weeks or months ago about that, I just wanted to give you a hint that it may not be non-annoying for some.

@Skillmonlikestopanswers.xyz I still don't think it's healthy and good for conversation to repeat the URL in every. single. chat. message.

@samcarter_is_at_topanswers.xyz said it's not spam (an information, right), but I call spam that I see is spam

@Skillmonlikestopanswers.xyz Just the part that some search index has listed: 252.000 occurences on TeX.SE. 14.100.000 occurences in the SE network. I may call it infested by spam. (German results, US english may differ)

@JosephWright OOH

@DavidCarlisle LOL

@JosephWright The one I saw from TUG2018?

@PauloCereda That's the one

@JosephWright Way better than mine. :)

@JosephWright new XPS would cost GBP 1,345.68 for me

@PauloCereda About the same here

@JosephWright wow

@PauloCereda dell.com/en-uk/shop/laptop-desktop-accessories-deals/new-xps-13/…

@JosephWright holy cow

Killer™ Wi-Fi 6 AX1650 (2 x 2) and Bluetooth 5.1

Oh no, a killer wi-fi

@PauloCereda Did you see the price of the UHD screen one?

@JosephWright oh let me see

@JosephWright wow

@PauloCereda There's a deal on dell.com/en-uk/shop/laptop-desktop-accessories-deals/new-xps-13/…, 14% off so actually £1,314.95 (if I had to replace my machine tomorrow, this is the one I'd get)

@StefanKottwitz I won't be mad if the people told me they find it annoying. As of now the only person who expressed that feeling is you, though you're doing so very prominently and vehemently it's still just one person.

@Skillmonlikestopanswers.xyz Sounds like a good good idea to make a survey on the meta site

@StefanKottwitz feel free to go ahead.

@Skillmonlikestopanswers.xyz Sure, possibly heavily downvoted by the spam site "buddies" :-)

@StefanKottwitz make it general, not only about TA and it shouldn't get any downvotes. (not that I think downvotes would be justified even if it was only about TA, but still keeping it general seems like the best way)

@Skillmonlikestopanswers.xyz, @StefanKottwitz Perhaps best if I post a question from a neutral POV?

> the spam site "buddies"

Someone linked to an interesting article recently english.stackexchange.com/questions/225523/…

@Skillmonlikestopanswers.xyz I had my experience when I tried to have a fair talk with the spam site users on another SE room. They starred and upvoted nearly every response of their spam site users, so putting all their points on the side bar making my comments and replies effectively invisible.

@samcarter_is_at_topanswers.xyz it's a quote of a term used by marmot, since he called us moderators "buddies" of his opponent. Nice that you listen!

@JosephWright Sounds good! Like if it's welcome to use a complete web site URL in the user name to transport it to every single question, answer, comment, and chat message where the user participates. No need to call users or the site by the name.

@StefanKottwitz still, if you want to be taken seriously and treated fairly by others, I advice you to refrain from using that rhethoric. Just because someone does something doesn't mean it's ok for you to do the same, especially in your role as a moderator.

SO employs statement about user names: "Whether they're advertising politics, a product, or an Area 51 proposal, there is nothing inherently wrong with using your display name as an advertisement for whatever you'd like to advertise [...]" from meta.stackoverflow.com/a/349167

@JosephWright sounds good! Thanks!

@Skillmonlikestopanswers.xyz My role as a moderator forced me suppress my free speech even when people lied and provocated, to calm it down somehow. Being bound by a role is not really a bonus.

@StefanKottwitz I never said your role would be a bonus, quite the contrary. The "especially in your role as a moderator" was meant as a "with the extra responsibility you were willing to take and take on a regular basis".

@Skillmonlikestopanswers.xyz Correct!

@samcarter_is_at_topanswers.xyz Finding a loophole to legally spam doesn't make it less annoying.

@StefanKottwitz starring is always a very biased thing. Take this room, I post a picture I find hilariously funny, others might find it boring. Then @PauloCereda posts something funny and gets 5 stars. Nothing bad about that. And iirc, your messages back then were unnecessarily provocating and emotionally charged and you didn't even try to stay objective in your rhethoric. Can you then blame the others for getting/being emotional as well.

@StefanKottwitz and @samcarter_is_at_topanswers.xyz, can we stop this discussion here for now? The meta post on tex.meta will (hopefully) settle things and we can appreciate a calm, peaceful and drama-free atmosphere here!

@JosephWright "ling background"? What does that mean?

@UlrikeFischer Must be a typo

@Skillmonlikestopanswers.xyz Sure, just don't forget to consider taking care of the issue on TA, that was the occasion to talk about it here, before the topic drifted off.

@StefanKottwitz, @Skillmonlikestopanswers.xyz, @samcarter_is_at_topanswers.xyz Will post something soon: got distracted looking over

@JosephWright I have a question about \hbox_to_zero:n. How does it differ from \hbox_overlap_right:n ? Does it perhaps turn into \hbox_overlap_left:n in an RTL context?

@AlexG \hbox_to_zero:n doesn't apply hss to the content, but \hbox_overlap_right:n does

@AlexG RTL semantics still not addressed for these things :(

@JosephWright Ah, thank you!

@JosephWright the linked meta answer explicitly includes "advertisement", not only "political statements". IMHO, it would be fair to change the wording in your question as it is explicitly about promotional material (advertisement).

@JosephWright It seems that \hbox_overlap_right:n and \hbox_overlap_left:n swap their meanings in RTL context. This is ok and is what I expect.

\documentclass{article}
\usepackage{expl3}

\TeXXeTstate=1
\ExplSyntaxOn

\begin{document}
\beginR
\vrule~width~1pt~height~2ex\hbox_overlap_right:n{This~box~overlaps~left.}\hbox_overlap_left:n{This~box~overlaps~right.}
\end{document}

So one programs for LTR, but it also works in RTL

@Skillmonlikestopanswers.xyz OK, I'll adjust

@AlexG My point was that we don't document this, there's no way to fix the direction, and one might argue that 'left' and 'right' are the wrong terms (they are LTR-centric)

@Skillmonlikestopanswers.xyz ducks and rabbits <3

@UlrikeFischer ooh ling

@JosephWright Perhaps l3build could catch on line xxxx when the line wraps: github.com/latex3/latex2e/commit/…

@PhelypeOleinik Pattern for the hit? %([^%)]+%)%s+line %d?

@JosephWright oh no

@JosephWright It may line-break in two places so perhaps on(%n%([^%)]+%)%s+| )line(%n%([^%)]+%)%s+| )%d+ (not sure if the right syntax for Lua)

@PhelypeOleinik Normalisation works on a per-line basis, so we are always looking for lines that have wrapped rather than those containing newline chars

@JosephWright Ah :/

@JosephWright So maybe %([^%)]+%)%s+(line )?%d+, but it might catch other (whatever)<spaces><number> it shouldn't

@PhelypeOleinik something like that, yes

@PhelypeOleinik I'll remind myself of the Lua and maybe adjust

@PhelypeOleinik BTW, upload has gone to CTAN

@JosephWright Maybe this one, so it doesn't match input files by accident (added a /): %([^%)/]+%)%s+(line )?%d+

Now we find out who is still using \etex_.../\pdftex_..., etc.

@JosephWright Thanks!

@JosephWright I think I'll wait to add \msg_term:n to the next dev release... (in the hook code, I mean)

@PhelypeOleinik Sure, got it

@PhelypeOleinik \msg_term:n was an old name: you'll need to be careful in the deprecation area

@JosephWright Oops. Missed one n :-)

@PhelypeOleinik Ah :)

@PhelypeOleinik Been reading Bruno's PRs?

@JosephWright Not as thoroughly as I'd like. But the peek analysis thing is impressive (and very handy, I think)

@PhelypeOleinik Probably we should schedule review on them, so we can discuss interfaces properly at the team meeting

@JosephWright Okay. Next one will probably be... calmer, after the -dev (or not :-)

@PhelypeOleinik Well yes, we likely need a couple of 'cycles'

ooh Captain Sir Thomas Moore

@PauloCereda Reading BBC news?

@JosephWright Auntie :)

@JosephWright ditch my logo... :(

@PauloCereda No! Build on your logo

@JosephWright Pollyannaism :)

@JosephWright ^^ :)

twitter.com/tds153/status/1282687551506190337

Hola hola, Someone here can tell me what you get by using lualatex --shell-escape test.tex over this file:

Update today and I get

@PabloGonzálezL You've got a stray format file

@JosephWright Always use tlmgr update --self --all --reinstall-forcibly-removed

How do I get out of the way?

@PabloGonzálezL Try kpsewhich --all --engine=luahbtex lualatex.fmt

@JosephWright I get the same thing :(

@PabloGonzálezL what do you get as output from Joseph's command?

@PabloGonzálezL and what is the date of this format?

and check also dvilualatex.fmt

@UlrikeFischer Using lualatex-dev works OK

@PabloGonzálezL well that is a different format. As Joseph wrote: the error indicates an old format somewhere. Is it lualatex or dvilualatex that breaks?

@UlrikeFischer It is lualatex the problem, the package auto-pst-pdf-lua generates the images correctly (it uses dvilualatex as far as I understand the code)

The big log pastebin.com/i9cfkSmP

@PabloGonzálezL no I meant: try a small hello world document with lualatex and dvilualatex. Which one errors?

@UlrikeFischer No problem with any of my other documents, it's just this specific one

@PabloGonzálezL did you try another document with dvilualatex????

@UlrikeFischer Yeah

@PabloGonzálezL your format is has a quite curious date: format=lualatex 2016.5.20. Did you try fmtutil-sys --byfmt=lualatex? Does it update the format?

@UlrikeFischer You saved my day fmtutil-sys --byfmt=lualatex together with fmtutil-user --byfmt=lualatex have solved the problem

No idea why this happened :(

@JosephWright I found this tex.stackexchange.com/a/125543/2388. And if I copy the setcolorspace part, followed by a 0.4 setcolor in the middle of a ps-file, I get after ps2pdf an entry in the /ColorSpace resource, a separation object (but with a missing new line before the endobj), and in the stream /R10 cs 0.399902 scn. Which means that is theoretically works, but the details are missing ;-).

@Skillmonlikestopanswers.xyz Regarding meta, I guess I was right: 6 downvotes already without even naming TA. Funnily this time inside spammers don't show up yet with their name to comment (since they have spam in their name?) Or can I expect 6 comments that explain what they think is wrong? Zero for now in that discussion.

@cis The Linux Mint homepage has a note about that:

> Tip: You can try to boot Linux Mint 64-bit on your computer. If it isn’t compatible, nothing bad will happen. You will just get an error message.

@UlrikeFischer Hmm

@StefanKottwitz I think discussion is best kept to the meta question

@JosephWright Sure, waiting for it.

@MarcelKrüger Oh, then I remember that I started the installation in "compatibility mode". I didn't know what that was pointing out, but sounded good.

@JosephWright it is a start. The main point that I was missing all the time is that the tintTransform must be a function, I was always trying with some arbitrary values and got errors.

@UlrikeFischer Ah, right, yes, but still tricky :)

@cis Especially on an older laptop that's probably normal, but it's been some times since I last used an installer to install linux, so I wouldn't know.

@cis Generally I would recommend against such compatibility modes unless necessary, they often disable some more efficient drivers and make everything slower.

@JosephWright yes, I will play around this is a bit. Someone who knows postscript will have to tell me how you can store the tintprocedure and the array in some variable/function.

@JosephWright oh that explains who gave me a downvote on an ancient answer:-)

@DavidCarlisle Seems this brought you on track for 666,666

@samcarter_is_at_topanswers.xyz that had been noticed (otherwise I'd have worked harder to rep cap so not get the -2 :-)

@DavidCarlisle :) What a considerate user to help you!

@samcarter_is_at_topanswers.xyz there's time to get more downvotes before then so it may yet come to nothing.

@JosephWright @DavidCarlisle How would you read "Petrine ministry"? "pea - try - ni"?

"Pontificate" seems easier, nevermind. :)

@PauloCereda not a word I know but I'd pronounce it as "pet reen" unless told otherwise

@DavidCarlisle Thanks, the Google lady suggested me "try" instead of "tree". But she also says "tee" instead of "ty" pography. :)

@PauloCereda Google :)

@PauloCereda three youtube pronunciation guides with three different versions. Typical English pronunciation rule "foreign looking word, just make something up" youtube.com/watch?v=nrSvLaFm0NQ youtube.com/watch?v=3aThQGGkVQQ youtube.com/watch?v=mL2ZIFOeDpg

@DavidCarlisle YouTube :)

@PabloGonzálezL which doc do you want, texdoc ma-cb-en.pdf gives something

@PabloGonzálezL don't you want -l not -f to get the number list

@DavidCarlisle Mmmm... I was hoping to get "cont-eni.pdf"

@PabloGonzálezL Sorry, no documentation found for "cont-eni.pdf".

@DavidCarlisle Mmmm... It is available at CTAN ctan.org/pkg/context

It's strange that I don't get it in my distribution (which is complete)

@PabloGonzálezL isnt that the mkii doc?

@DavidCarlisle I can't find it in my system :( ... Can you tell me the path of that file in your distribution?

@PabloGonzálezL oh I see it's excluded from texlive due to the licence, look at the first page on the ctan copy it prohibts distribution

@PabloGonzálezL but it's from 2001 and set with pdftex so it's the old manual

@DavidCarlisle I see, you are right, it is a bit old document, the texdoc ma-cb-en.pdf file is what I am looking for :) [It should come out first when using texdoc context]

@PabloGonzálezL request that on texlive list and they may add a special case rule so that happens

@DavidCarlisle I think texdoc has its own list or not?

@PabloGonzálezL possibly.

@DavidCarlisle It's almost as useful as texdoc latex :-)

@UlrikeFischer ^^ better than silly blub examples

@cis Sounds very long. I don't have experience with mint, but installing fedora on a similar old laptop took me maybe half an hour

@cis What does your screen show at the moment?

@cis fedora-mate... always works better on older equipment... like my old DELL :)

@DavidCarlisle The etexcmds package is functional today ... or it is not necessary to protect macros that collide with ConText

@PabloGonzálezL No need to restrict this statement to older equipment, fedora works better on everything :)

@samcarter_is_at_topanswers.xyz That is an absolute truth! Linux worktex 5.7.8-200.fc32.x86_64

@DavidCarlisle fair is foul and foul is fair ...

@UlrikeFischer I increased the allowed form size by a factor of 100 not sure how generous I should be, but something isn't right the table of contents isn't showing the No file docuemnt.toc in the log is supposed to trigger a re-run, I need to test with a smaller example

In \includegraphics, I'm not sure how the trim arguments work.

Are those coordinates for the image as they appear on the page?

@PabloGonzálezL I doubt it has beentested against context in a long time

@DavidCarlisle how many pages does it create?

@UlrikeFischer see the toolbar in the image:-) 41

locally I get Output written on macbeth.pdf (41 pages, 193424 bytes).

@DavidCarlisle ah ;-). That isn't so much, if it is mainly text.

@FaheemMitha they are relative to the bounding box of the image

@DavidCarlisle It sounds like the image itself wouldn't get scaled up, then.

@UlrikeFischer yes the author appears to be very conservative he's actually written individual words, no recursive loops or lipsum calls to generate the text.

@FaheemMitha trim doesn't scale anything it's just for trimming off the edges, as with just cutting the border off a print with scissors

@DavidCarlisle Ok. I want the trimmed image to be scaled to fit the page. Is there a graceful way of doing this within the confines of \includegraphics?

@FaheemMitha width=\textwidth

@DavidCarlisle Oh.

@DavidCarlisle Is keepaspectratio enabled by default?

@PauloCereda no, well yes but with value false

@DavidCarlisle got it :)

@DavidCarlisle Thank you, it's clearer to me

@UlrikeFischer oh it's hyperref that broke it

With

\documentclass{book}


%\usepackage[colorlinks, linkcolor=blue]{hyperref}

\begin{document}


\tableofcontents

\section{aaa}
aa

\section{bbb}
ccc

\end{document}

after the first run you get No file cn.toc but if you delete everything and start again but with hyperref uncommented then you do not get that warning

@UlrikeFischer I had better look also for (rerunfilecheck) Rerun to get outlines right

@StefanKottwitz is that meant to be provocative or just sounds like it by accident? Also, I for one haven't voted at all and don't want to express my own views on this as they are heavily biased and I want to see what a fair conclusion of the majority of the frequenters of this site looks like.

@DavidCarlisle so you can blame me? ;-)

@UlrikeFischer ooh

@Skillmonlikestopanswers.xyz Just openly sharing with you what I saw there, since we talked. The for meta uncommon pattern looks like other people don't want to see such a fair conclusion of users but to push it in a desired direction. Still there's the possibility that 6 non-biased users downvote, but I don't understand yet which part they might mean. If it's sock puppets, they won't tell.

@UlrikeFischer naturally, been out for a bit I'll have a look what happens, but I should catch the rerunfilecheck messages anyway

@Skillmonlikestopanswers.xyz And this time it's not randomly starring (right sidebar) of funny pictures or jokes or important news, but it's picking and starring pro-topanswers messages out of a comprehensive conversation. Interpret any way you think, it's just the same behavior I noticed when I talked with them in their duckoverflow chat room months ago.

@StefanKottwitz neither do I, that's an inherent with such voting systems.

@DavidCarlisle learnmacbeth.org?

@Skillmonlikestopanswers.xyz I appreciate talking with you, and by the way, the original issue is gone, thanks!

@PhelypeOleinik to latex or context, that is the question

@DavidCarlisle HaM \let :-)

(this is as bad as my jokes can get :-)

@StefanKottwitz you mean currently in this chat? The messages I wrote which got starred were all those messages in which I either stated that I try to find an objective consensus, or the one in which I tried to end the discussion in order to wait for the consensus on meta, nothing pro-topanswers as far as I can tell.

@PhelypeOleinik that one is available too if you wish

@PhelypeOleinik Germans take Goe\the:-)

@StefanKottwitz :)

@StefanKottwitz Just for comparison: similar messages in stackexchange chat that spread lies about other users are not gone

@samcarter_is_at_topanswers.xyz Like those I saw in the duckoverflow chat room? Flag what you mean (again if needed) as this is just too vague, I simply don't know which date and where to search for it.

@UlrikeFischer ouch I'd never noticed hyperref does this, it discards an existing toc and suppresses the message that it isn't there. :(

% If hyperref was used last time, do nothing. If it was not used,
% or an old version of hyperref was used, don't use that TOC at all
% but generate a warning. Not ideal, but better than failing
% with pre-5.0 hyperref TOCs.

@samcarter_is_at_topanswers.xyz By the way, when flagging please justify/explain so mods can understand, in some cases an unclear flag may be dismissed because mods may simply not understand why, but just got a flag wit a claim, in rare cases. (such flags sometimes stay long in the queue because nobody can decide). Just, the clearer the better, in short.

Mods are happy to serve for clear things.

@DavidCarlisle I know the warning but never checked why and if it is still needed - but probably we need always 4 arguments for contentsline to get rid of it.

@UlrikeFischer adding a test for rerunfilecheck messages seems like a good idea anyway so it's fixed at latexcgi but another hyperref quirk to clean up one day:-)

@UlrikeFischer and @DavidCarlisle -- The number of arguments for \tocline (with or without hyperref) was a real pain for AMS. What problem (other the nuisance of figuring out where to do it} would result from simply adding an empty {} at the end of the input?

@DavidCarlisle what could be the problem? Simply adding an empty argument in the \addcontentsline definition seems to work fine?

@UlrikeFischer i am not sure, the comment doesn't say it fails with a standard definition but a pre 5.0 hyperref one so I'd have to dig out a hyperref 4 and see what it was doing

@barbarabeeton at some point we will probably want hyperlinking to be built in but getting the timing right is tricky but this particular comment seems to be worrying about running a 25 year old file while keeping the .aux and .toc from the old system and avoiding errors

@DavidCarlisle No, I meant why can't the latex kernel add an empty brace pair:

\def\addcontentsline#1#2#3{%
  \addtocontents{#1}{\protect\contentsline{#2}{#3}{\thepage}{}%
                     \protected@file@percent}}

@UlrikeFischer I think we tried it and some patch in koma or memoir or some random university thesis class from 1984 broke,

@UlrikeFischer but we have got better at getting changes in so maybe we could do this ...

@UlrikeFischer -- Yes, that's the sort of adjustment I was thinking of. I think that would work fine in the AMS classes. (But do realize that memoir and KOMA might behave differently).

@DavidCarlisle I think we should try.

@UlrikeFischer yes especially with \protected@file@percent the code has already changed in the not so distant past so old patches expecting exactly the 2.09 definition are already broken

@DavidCarlisle it would be even better, if there where some dummy command there. Then hyperref could simply define it, instead of patching \addtocontents. Well looking at the patches, it isn't so easy ;-(

@PabloGonzálezL @PauloCereda What kind of cuia do I need for that? :)

@yo' Did you see my question a few days ago about removing non-aux extra files from the cache on Overleaf?

@AlanMunn One made of Porongo. Anything else makes the chimarrão taste weird (@PauloCereda wouldn't know; he's too far north :-)

@PhelypeOleinik True.

@PhelypeOleinik But I was asking specifically about fedora-mate. Maybe it's different. :)

@AlanMunn A laptop made of Porongo? :-)

@PhelypeOleinik New from Dell, the Luddite :)

spot the difference:

%!TEX program = pdflatex
\documentclass{book}

\begin{document}
zzz
\end{document}

%!TEX program = pdflatex
\documentclass{book}

\begin{document}
zzz
\end{document}

Actually I'm not sure the difference shows up but one is accepted by overleaf and one isn't most likely because of

$ file cn.tex
cn.tex: LaTeX 2e document, ASCII text

$ file cn.tex
cn.tex: PostScript document text

We can blame @yo' :-)

@DavidCarlisle what is the difference?

@UlrikeFischer newline at the start %! as first two bytes is a unix magic marker for PostScript....

@UlrikeFischer finally realised why sometimes the open in Overleaf button does that^^^^

@DavidCarlisle I never really tried to open in overleaf, the other button is so much faster ;-)

@UlrikeFischer The other button works without registration

@StefanKottwitz yes although the observed differences are probably due to other things like not opening in a new tab which makes it seem quicker even if it isn't and a lighter weight setup at the server, the machine itself as you observed the other day isn't that powerful or quick:-)

@DavidCarlisle Do you monitor the load over time?

I mean like CPU or memory, not hits.

@StefanKottwitz no not a lot of point really at the moment as it runs at around ~30 documents per day for a second or so each so load is 0 almost always.... I'd need to keep more of an eye on it if it became popular...

@DavidCarlisle Yes, for later. Not sure what is a good lightweight tool to track that.