@DavidCarlisle Go figure. :)

@Bob @DavidCarlisle @UlrikeFischer The concept of "floating" images and tables boggles my mind. Why would you even want your things not in the place where you talk about them?? (can you tell that I dislike endnotes?)

@Plergux I dislike endnotes, too

@Plergux IMO some books and articles look better when certain kind of content goes in a fixed position in the page. Think about those one-page tables or images for which a "put me just here" would make presentation à la MSWord (read "incredibly awful")

@Plergux Sometimes the material is purely supplemental. I regularly read papers about CS with new algorithms being introduced. And it's really nice that they have some table with performance measurements but the really interesting part is the text evaluating the results. Having the table not disturb the reading flow is beneficial to me because I focus on the relevant part and look at the table (which is close due to floating) afterwards.

@CarLaTeX The only thing that annoys me more than endnotes is repeated footnotes that take over third of the page. If it's that damn important put it in the main text! :p

@Plergux because pages have this annoying habit of having a fixed size.

@DavidCarlisle wasn't there some toilet paper class that fixed that (at least in one axis)?

@Skillmon probably although there are limits on how long you can make a page in pdf

@JairoA.delRio Well, of course there are exceptions. A table that fills a whole page needs to be placed on the next available full page, but several times I tried to do that with [ht] in my thesis they would jump to the end of the chapter (which in some cases was over fifty pages away).

@DavidCarlisle and there are limits in TeX (which might be smaller than those of PDF, I don't know)

@DavidCarlisle And? I don't get it.

@Plergux that is not surprising [ht] makes it more likely the float goes to the end of the document. If you get what you ask for, don't blame latex:-)

@TeXnician Then you put it in the supplement (appendix). :p

@Plergux point is, that manual placement of figures etc. can be better, but due to the restricted paper size, automatically placing them where you put them in the sources most likely doesn't equal best manual placement. Floats give good results on average, and you can later go the extra mile and place individual floats manually to get better placement.

@Plergux Nah, computer scientists love their tables and performance figures because they want to show off :D

@Plergux if you have a table that takes up say 30% of the page height and the "natural" place to reference that table comes 75% down the page, the overwhelmingly standard thing to do, since before printing began would not be to leave the page short but to let the text carry on but reference the table at the top of this page or the next, that is what floating does,

@TeXnician No computer scientist here. Take a look at expkv :)

@DavidCarlisle well, that's no good. Who else can I blame then?

@Plergux @UlrikeFischer of course

@Skillmon Oh, that shouldn't be understood to be exclusive to CS people. It's just that my story above was about CS papers where people love to show off speed gains using their algorithms, preventing to put things like performance tables into the appendix. With expkv you are far ahead of many computer scientists because you actually publish benchmark code and not only show off figures in a table :D

@Plergux @Plergux but if you have a large figure more than 70% of the page (this fraction being a user settable amount) latex won't place it on a page with a few lines of text from the main document flow, it places it on a page on its own. but if you use [ht] the main thing is that this means not p ie not on a page on its own. If you stop it going on its own page and it does not fit on a text page, it is not allowed anywhere so it, and all following figures get pushed off to the end

@TeXnician agreed, benchmarks which don't show how they were done are pretty meaningless. I guess I could always find some edge-case in which my algorithm is faster than anyone else's...

@Skillmon It's probably just me then cause the thought of letting the computer decide where to put my stuff curdles my blood. :p My car has this thing where it displays on the dashboard whether I should change gears and it grinds MY gears no end :p

@TeXnician :D

@Plergux there is a good answer on this network regarding floats. Let me dig real quick.

@Skillmon Frank's opus?

@DavidCarlisle @Plergux yep: tex.stackexchange.com/a/39020/117050

@DavidCarlisle But this computer stuff, by design, lets you squish and stretch to your hearts content! If there is only room for 95% of the table just change the table.

@DavidCarlisle ah of course. silly of me. :p

@DavidCarlisle ah, then possibly I have misunderstood "p" :p

@Plergux well apart from the case that you talk in more than one place about them: even if you dislike endnotes you seem to be fine with footnotes floating to the bottom of the page, so where is the problem if a figure floats there too or to the top?

@UlrikeFischer Because a footnote is unimportant stuff that people can read if they want to (and don't have to turn to the end to do so), but figures and tables is important stuff I want them to read now. :p

@Skillmon ooh... thanks. :D

@Plergux humans can do a better job, if you use [H] and re-write the text and move the table in the source so it comes out right. But that isn't what you see happening. latex moves figures to avoid bad white space at page breaks, people use [H] then complain they get ugly large spaces. [H] by design stops latex avoiding ugly spaces, the intention was that the author takes action, eg by writing some more words to fill the gap. That isn't the answer people seem to want though.

Good morning, everyone!

Quick question, does anyone have any thoughts on the issue discussed in the comments here? tex.stackexchange.com/questions/587979/…

Specifically, egreg -- whose expertise in all things TeX is of course beyond question -- assures me that \documentclass{article}\usepackage{textgreek}\begin{document}\~\textnu\end{document} cannot possibly work, i.e. produce a tilde above a Greek letter nu, yet it does for me.

@chsk I get ^ with pdfltex (texlive 2021) what system did you use and what output did you get?

@Plergux yes sometimes there is such a design. I had once a document with small sections and a plot in everyone along with a text describing it. The plots couldn't float but one had to adapt the text very carefully to fill in the spaces. But this isn't always the case. You also often have a two page text describing some table or figure, and the figure can be everywhere near this text. And then it helps if you simply let latex place it "somewhere around here".

@DavidCarlisle MiKTeX/xelatex.

@chsk ah well xetex is completely different:-)

@DavidCarlisle Exactly. :)

@chsk beat you to it:-)

Great minds think alike. ;)

@DavidCarlisle Well, users would be on a normal curve, wouldn't they? Which means 45% are below average.

@DavidCarlisle In any case, that's interesting then. How come that this works in xetex but not pdftex?

@chsk but textgreek with xelatex doesn't sounds right, it uses LGR encoding.

@UlrikeFischer What does that mean?

@chsk in classic tex to get greek you have to do "comoplicated stuff" change fonts change encodings work out where nu is encoded etc, in a unicode tex textnu is just a somple macro to get ν

@chsk you are using an unicode engine but a non-unicode font.

quack

@UlrikeFischer In many of these cases I have been told by my thesis advisors that I should have a figure dedicated paragraph (short description), then figure, and then further discussion of said figure (when first appearing). So it's also a problem of what you're being asked to do as opposed to what the architects of LaTeX have deemed to be "normal behaviour" for a figure.

@chsk with xelatex the accent is a combining accent: \add@unicode@accent {"0303}{\textnu }

@PauloCereda ΠΡΩΙΝΟ ΓΕΥΜΑ

@UlrikeFischer I see, thank you.

@DavidCarlisle oh no

@JosephWright happy St. Joseph's day

@Plergux no as I said to bob last night, latex does not assume that you want images to move, the default behaviour is for them to be placed like a letter they go exactly where you place them, subject to line breaking. But if you want to allow things to be taken out of the document flow and inserted somewhere then it provides figure wrapper which does that. That is the only purppse of that environment, it is unrelated to image insertion, it is just about moving its (arbitrary) content

Thanks David and Ulrike. Have a nice day and a good weekend! :)

@Plergux You'd like Edward Tufte's approach: he doesn't use callouts or floats at all. But then he controls everything about his books so the text is 'just right'

@Plergux As a chemist, this comes up a lot: we favour 'here' for schemes ('chemical equations') as they read to some extent like mathematical equations. But they are also quite big, so it's not necessarily easy

@JosephWright so Chemistry is pretty much German with symbols? :)

@Plergux but if you read Frank description you can see that the normal behaviour can be easily adjusted. E.g. you can simply add \def\fps@figure{!htbp} and then latex will try to keep your figures in place. (But I personally greatly prefer if figures are at the top or botton of a page as I find them disruptive in the middle of a page.)

@PauloCereda and explosions

@DavidCarlisle ooh

@PauloCereda and sometimes smelly.

@Skillmon ooh

@PauloCereda counter-ooh

@Skillmon <3

@PauloCereda ooh, a carrot! Here, have two, too: <3 ε>

@Skillmon as you are here we are planning to go with the current rawoptionslist code for the upcomming -dev release so I'm planning to close your open issue on that. it is in latex-dev so it can still be tweaked if necessary before we move it to the main release, so feel free to open a new issue if things can be improved after you have tried out -dev for a bit

@DavidCarlisle All right, I can see that (although most tutorials I've gone through simply say "put tables in 'table' and put images in 'figure') but is there no alternative environment to put images in that doesn't float? Do I just resign myself to scrapping the extra environments altogether and just using captionof?

@JosephWright That is very appealing. :p I'll have to read up on that :p

@Plergux yes minipage and captionof is fine, actually using figiure and [H] (which is almost the same thing) is also fine, it's just more weird than people seem to think. It's like getting a non moving seat by wheel clamping a ferrari when you could just use a seat.

@Furdzik good to see another monochrome human, makes a change from assorted animals in chat:-)

@DavidCarlisle ok, thanks for letting me know :)

@UlrikeFischer I am exactly the opposite! I prefer to transition between pages with text rather than having random pictures jumping in XD

@Plergux A wild picture appears

PLERGUX uses GTFO

It's super effective!

@PauloCereda XD

> The US Government Finally Gets Serious About IoT Security

^^ ooh Island of TeX... oh wait

@AlanMunn a linguistic intermezzo :)

@DavidCarlisle ok. I will look into that. It's probably too late to try to implement it in my thesis (though I might be tempted to scare my advisor with it XD).

@yo' LOL

@DavidCarlisle but for a beginner, it's easier to get consistent spaces around figures using [H], I guess.

@yo' Is Loud-Ness pro-independence? :)

@yo' poor ness is ill!

ooh

I wub you guys

@JosephWright or @DavidCarlisle regarding tex.stackexchange.com/questions/588022/… is there a place the complete code documentation of expl3 is provided or does one have to build it by oneself?

@JairoA.delRio who knows, but obviously it's Scottish ;)

@yo' ooh

@Skillmon on there is interface3 and expl3 (which is probably more tahn enough in taht context) but for sources you need to build your own doc

@PauloCereda nessy taking lunch home

@DavidCarlisle oh no

@DavidCarlisle Uber Eats in Scotland? :)

Oh no

@Skillmon source3?

@JosephWright ooh sourcetreeapp.com oh wait source three not tree :)

@PauloCereda Biadh-maidne

@JairoA.delRio oh no

@DavidCarlisle ^^

@JosephWright ooh, I didn't know it was distributed in TL... :)

@Skillmon just shows, you should believe @JosephWright not me.

@DavidCarlisle I'm sure we'll find a way to blame @UlrikeFischer that your information wasn't up to date!

@PauloCereda You chose the IoT name wisely ;-)

@PauloCereda Our IT Servicedesk had the name IS before it was changed a few years ago for reasons ;-)

@StefanKottwitz <3

@StefanKottwitz oopsie :)

@PauloCereda yep, oopsieness.

@StefanKottwitz :)

That's curious... I have a document that produces different font adjustments when I use XeTeX instead of LuaTeX...

HarfBuzz thingy?

:)

@PauloCereda Because they are not the same program?

@barbarabeeton Oh, Blueno is \Huge! It looks really funny, too bad it had to retire!

@DavidCarlisle okay, that ends the subject.

@PauloCereda Obviously the LuaTeX document looks much better?

@MarcelKrüger it does :)

@PauloCereda If you send me the document I can tell you what's going on.

@MarcelKrüger thanks, but it's not too important, it was more of a curiosity. :) Unless you are curious too. :)

@PauloCereda I thought it was a rather thorough analysis.

@PauloCereda I have a feeling that it's caused by the calt feature being disabled by default, so it's probably a known difference.

@Plergux you're right!

@MarcelKrüger Ah then hold on, a document is on its way!

@MarcelKrüger sent!

@PauloCereda Shockingly, my crystal ball was wrong. The difference in the documents sees to be mostly caused by microtype having much better LuaTeX support.

@MarcelKrüger ooh :)

@PauloCereda I refer you to my earlier answer.

@DavidCarlisle I refer you to my earlier reply as well. :)

@MarcelKrüger thanks, Marcel!

@MarcelKrüger you are right tabular rules would be a problem. Do you have an idea then for a suitable end-marker for fbox, something like an lua-xspace?

@UlrikeFischer ?

@Plergux quack

@UlrikeFischer You could change the glyph code to scan not only look at the next node but also skip rules and stuff, but then we still have to deal with the boxing level. It might work to add a marker node to lists which end with a glyph such that hboxes with this marker could be handled like glyphs.

@DavidCarlisle the luacode to insert real spaces doesn't recognize a space after an \fbox and so \fbox{box} and is exported as boxand. I'm looking for something that marks an potential space there.

@UlrikeFischer The code could also be extended to also look back from glyph nodes. Then glue before glyphs would also get marked, which at least would handle the common case where you don't have two \fboxes next to each other.

@MarcelKrüger would that catch things like the paragraph indentation? I'm not sure if I would want a space there.

@UlrikeFischer The paragraph indentation is a empty box, not a skip. Glue which ends up at the start of the line is discarded. (Except for \leftskip, but that isn't present before linebreaking when the attributes are set.)

@MarcelKrüger which (reasonable sized) horizontal glue would you not want to mark?

@DavidCarlisle The tabular issue was about the glue between a vertical rule of a tabular and the cell content. It might be roughly space sized and probably shouldn't get marked.

@MarcelKrüger ah, I don't suppose you allow yourself to change the macro layer so those spaces are differently marked when added?

@MarcelKrüger but if I have a table row that is red&blue&green\\ wouldn't I want that to cut and paste as red blue green with spaces coming from "somewhere" ?

But IMO the bigger issue is deciding what "reasonably sized" means here. Also it's a big unclear how unreasonably sized rules should be handled. "abc\hfill def" should probably set a space in the middle, but some code which uses similarly sized glue to position the footer probably shouldn't have random glue in the middle.

@MarcelKrüger well yes I chose the wording carefully to avoid specifying that...

@DavidCarlisle It's Ulrike's code, so she will have to decide how much adaption of the macro layer can be assumed. I think I would favor only marking spaceskip subtype glue + explicitly marked glue.

@DavidCarlisle Probably, but that should be independent from the existence of vertical rules, so most likely the tabskip.

@MarcelKrüger except tabskip is usually 0 in latex with the column space being a normal \hspace added within the cell box

@DavidCarlisle yes, but this should happen as all cells are marked as tabular cells, not with spaces.

@MarcelKrüger looking back seems to work in my case. But now I wonder why my code looks only for kern followed by glue, but doesn't insert a space for text\kern1cm in ;-(.

@UlrikeFischer IMO that's the right thing to do. Using kern instead of glue indicates that it probably isn't supposed to be "space-like", so it shouldn't get a space glyph. Just pass on the blame to whoever wrote a kern there.

@MarcelKrüger well blaming me sounds sensible ;-). How can I test if the previous glue has already the attribute? Then I prefer not to overwrite it in case there was a font change.

@UlrikeFischer node.has_attribute?

@MarcelKrüger yes, thanks that should do it -- but I have to make lunch now first ;-)

@yo' Cute Ness. :)

@PauloCereda baaa

@Plergux hi <3

@PauloCereda ho <3

@Plergux ooh

@Plergux -- Should I really believe that you don't know about [p] for full-page tables or figures? [ht] just don't allow full-page objects at all, hence end of chapter.

\make{lemon}

@MarkGiraffe \make{lemonade}!

@barbarabeeton I knew about "p" but I always understood it as "separate page for images" in the sense that it would just shove all images on separate pages automatically allocated to hold images as need would be. So if I had three images throughout the text with [p] there would in the end be for example three pages of text and one image page with the three images.

@Plergux yes if you have [p] that is more or less what it means as that is not t, not b and not h so [p] on its own prevents the floats being placed with text

@Plergux the positioning specifiers work as independent flags, setting the p flag means float pages are allowed on top of the other flags you set, so htbp means place here, if not possible top or bottom, if not possible on a float-page.

@Plergux and the order in which the possibilities are checked is always the same, so h always has highest priority if you allowed it.

@DavidCarlisle ok, so I'm not misunderstanding it.

@Skillmon yeah, but having all the flags is like driving an automatic shift car. you start the damn thing and it just goes into whatever gear it damn well pleases and you have to be stomping on the breaks the whole time. :p #rage :p

@MarcelKrüger My husband claims that „Kerns and Gallowglasses“ were done away with by Macbeth and Banquo. ;-)

@UlrikeFischer ooh the Scottish play

@Plergux it is OK to restrict the options but in that case you should set suitable numeric constraints. If you have \topfraction as .7 and then use [t] any float that is bigger than 70% of the page height (and all following floats) is forced to the end of the document unless flushed out with \clearpage. If you use [tp] then such a float will be placed on the next page. So there are very few cases where omitting p from the option improves things

@DavidCarlisle But that would mean that if I had 0.1 text, 0.8 float, 0.1 text I would end up having 0.8 of the text from the next page jump up and the 0.8 float move to the next page. It might be a style thing but I'm not seeing how that would improve anything.

@Plergux well if you actually want a specification that any float bigger than 70% of the page may not be positioned that is fine. (of course noe of this applies to H if you want to do full manual contriol you can always win. But I don't think people who routinely do \begin{figure}[t] or \begin{figure}[h] and then complain about latex float positioning realise that they are specifying that any large floats, may not be positioned anywhere and will go to end of document.

@Plergux Bug frank about 'Alice'

@DavidCarlisle Probably not. Certainly not I. And don't remember any of the tutorials that I've seen or read have ever said "floats are for floating things, if you don't want things to float don't use them". It's always "you put 'table' in 'table' and 'image' in 'figure'".

@JosephWright eh?

@Plergux Frank has written a couple of clever approaches to floats that go well beyond the current mechanism. The first, xor, is in the 'historical' part of the L3 repo, but we will likely not go that way. The second, 'Alice', is LuaTeX-based and very very clever, but FMi is cautious about making it public

@Plergux It's called 'Alice' because Alice in Wonderland is a real challenge for automatic typesetting

@Plergux Yes, and this advice haunts us perpetually. :)

@AlanMunn ConTeXt ....

@JosephWright Not planning to \start using it any time soon. :)

@AlanMunn \dodostartindeed

@JosephWright We'd best \stop before this gets out of hand.

@AlanMunn Oh no!

@JosephWright Ooh

@JosephWright Ah, I see. :) This debate could for certain turn out to become a long "tail" :p

Apparently the picture effect works on other sites too Where does the use of “deck” to mean “set of slides” come from?. This is a truly terrible answer but it has 65 upvotes.

ooh boughs of holly

@AlanMunn we could lure David to write a picture mode answer and see how it goes. :)

ooh there's a plant here

Hi plant person

Hi

Welcome to the chat!

Thanks! I'm quite new around here haha

Make yourself at home, we are a friendly bunch. :)

@PauloCereda When we're not being mean.

@AlanMunn ooh :)

@AlanMunn But you're the caring kind of mean. :p There is an Icelandic saying that says "A true friend is the one who tells you your faults" :p

@AlanMunn I also said bunch, not brunch (@DavidCarlisle). :)

@PauloCereda \puts away the croissants... Dangit! :p

@Plergux In Portuguese we have something along the lines of "A true friend knows everything about you and still insists on being your friend" :)

@Plergux oh no

coin coin

It's quack in French

@PauloCereda :D Those are the best friends :D

@Plergux I'm the kind of friend who will help you hide a dead body, but if you betray me, just remember: I know how to hide a dead body. :)

@PauloCereda XD

@PauloCereda Speaking of Portuguese, is there a reason why the main page on learnlatex hasn't been translated yet?

@AlanMunn oh, no idea...

@PauloCereda I thought maybe a battle about 3rd person clitics. :)

@AlanMunn ooh :D

@Plergux see what I mean? tex.stackexchange.com/questions/588058/…

@AlanMunn Some paragraphs like “Introduction” and “How it works” were added after the bulk of the text was translated, and the bloke who translated still didn't find the courage to diff the repository and update :)

@DavidCarlisle Yes. It's repeated "It doesn't do what it's supposed to" when "supposed to" is assumed to be different from what it actually does (which is what the user has essentially told it to do). I will admit my own guilt in this context (or luatext if you prefer :p) and hereby solemnly swear to try to understand stuff better before complaining. :p

@Plergux -- I've tackled placement of floats (not uncommon in TUGboat), and have a mental list of tactics to deal with them. Some have been posted here, and some even published in TUGboat. I think it's time to write them down in one place, probably in the form of an ordered checklist. If you'd like a copy (it may take a week or more to get it in a form that doesn't contradict itself), send me an email note (bnb at tug.org).

@Plergux Yes, this is a common problem, thinking that computers understand what we want, and then complaining when they do what we ask. :)

@Plergux don't mess with your thesis text based on anything anyone says here though! (especially things I say)

@barbarabeeton Cool :D I'll do that :)

@Plergux -- Around here, "puts away" (with respect to edibles) can often mean "consume". So your statement is ambiguous.

@AlanMunn True. The harsh truth is that the computer is only as smart as its user. I may have a slight advantage in this regard since I have never trusted the damn things. Usually when people start the whole Mac/Linux/Windows argument regarding stability my standard reply is "It's a computer, its going to fuck you up!" :p But then again, that might also just be me (I tend to do my utmost best to break things. :p)

@DavidCarlisle like finished your thesis?

@DavidCarlisle Too late. You're all gonna be in the credits :p

@Plergux WE ARE FAMOUS

@PhelypeOleinik get @AlanMunn to supply the text

@barbarabeeton I see. Well, maybe the "dangit" was an exclamation of regret of having eaten the croissants when it wasn't really eating time :p

@AlanMunn ^^^^^^

@DavidCarlisle Sounds like a plan!

@PauloCereda who would say that? It would be cruel to prod someome already stressed out with thesis deadlines

My father's day gift from my family

@PauloCereda :p If I remember correctly the line is something like "I would like to thank these people for helping me with typesetting (and there was DEFINITELY NO procrastination involved!)." :p

@Plergux ooh saved by the bell

@Plergux ooh

@DavidCarlisle ooh

@Rmano ooh

@Plergux -- Never trust a computer. My favorite mishap was getting an "out of order" condition on the output of a utility sort (on a mainframe). After patiently examining an inch and a half of core dump on greenbar, it was determined that somehow a bit had been flipped in a sort field. Cosmic ray? Someone plugging in a coffee pot? Never figured that out. But it wasn't the programmer!

@barbarabeeton Neutrons

Neutrons from space ....

Where's that programming with butterflies XKCD comic strip?

@JosephWright ooh that History Channel guy

The crazy guy

@UlrikeFischer What would be the expl3 for the following code?

      \pdfcompresslevel=0
      \immediate\pdfobj stream attr {/Type /Metadata /Subtype /XML} file{#1.xmpi}
      \pdfcatalog{/Metadata \the\pdflastobj\space 0 R}

@egreg with or without the pdfmanagement?

@UlrikeFischer I don't know what pdfmanagement is, sorry

@UlrikeFischer I found \pdf_object_write:nn which should be the guy.

@UlrikeFischer It's the xmpincl package.

@egreg yes, that is for the pdfobj. But to write into the catalog (the \pdfcatalog thing), you actually need our new code, as this must be managed: it would clash if two packages both try to add an entry.

@UlrikeFischer So it's basically unsupported in the current kernel/expl3?

@egreg More-or-less: this is the whole point of needed PDF resource management in the kernel

@JosephWright Does this mean that also using directly \pdfcatalog can be faulty?

@egreg See @UlrikeFischer's point: it only works if there is a single package trying it

@JosephWright I guess the problem is /Metadata and another package might want to add to the same part of the catalog?

@egreg yes, if you do this:

and then look into the pdf you see in the catalog

/Metadata /Nonsense/Metadata /MoreNonsense

And such duplicated keys are not allowed/faulty. The pdf viewer ignores then one.

@UlrikeFischer Is adding to the catalog necessary?

@egreg yes, that is the reference to the Metadata, the viewer can't find them if you don't have an entry in the catalog. (but naturally you normally don't write Nonsense)

@egreg with the new pdfmanagement, we will avoid such duplicates. So if two packages try to add metadata, only the last will have an entry in the catalog and so will will. You can try it:

\begin{filecontents}{whateverA.xmpi}
xxx
\end{filecontents}
\begin{filecontents}{whateverB.xmpi}
yyy
\end{filecontents}

\RequirePackage{pdfmanagement-testphase}
\DeclareDocumentMetadata{uncompress}
\documentclass{article}
\ExplSyntaxOn
  \pdf_object_unnamed_write:nn{fstream}{{/Type~/Metadata~/Subtype~/XML}{whateverA.xmpi}}%
  \pdfmanagement_add:nnx {Catalog} {Metadata}{\pdf_object_ref_last:}
  \pdf_object_unnamed_write:nn{fstream}{{/Type~/Metadata~/Subtype~/XML}{whateverB.xmpi}}%
  \pdfmanagement_add:nnx {Catalog} {Metadata}{\pdf_object_ref_last:}

The catalog then looks like this:

<<
/Type /Catalog
/Pages 11 0 R
/Lang (en-US)/Metadata 6 0 R
>>

@UlrikeFischer Then, for the time being, I can propose

    \pdf_object_unnamed_write:nx{fstream}{{/Type~/Metadata~/Subtype~/XML}{#1.xmpi}}
    \pdfcatalog{/Metadata~\the\pdflastobj\space 0~R}

@UlrikeFischer And let the user hope that nothing will add /Metadata to the catalog

@egreg yes, but along with the warning that this is incompatible with future latex development. I will at some time to to get the hyperxmp maintainer to extends his package to support such additions too so that one can drop xmpincl.

@egreg basically that means, the user shouldn't use hyperxmp or pdfx.

OK, I'll post for the questioner the expl3 version of xmpincl, with the warnings.

For all you cold ducks out there: facebook.com/687389255/posts/10160541628704256

@UlrikeFischer By the way, the doc for \pdf_object_unnamed_write:nn has the order of the arguments reversed: it says “Two brace groups: ⟨file name⟩ and ⟨file content⟩”

@egreg I should better check if the documentation or my code is wrong ;-)

@UlrikeFischer No, the code works, so it's the doc that is reversed.

@egreg yes, and imho <file content> isn't good anyway. It should probably be something like stream dictionary content. <-- @JosephWright ^^^

@UlrikeFischer Ah, right: we should change that

@JosephWright wasn't there somewhere a function where you get the csname as string without the backslash?

@UlrikeFischer \cs_to_str:N?

hello, someone have an idea about this:tex.stackexchange.com/questions/587984/…

@barbarabeeton ooh

@barbarabeeton hear hear!

@PauloCereda XD

@JosephWright yes that's what I meant.

@PhelypeOleinik O meu portugues é só de ler.

@AlanMunn Obviamente você não é um lingüista de verdade

\start{talking}

@DavidCarlisle Tehno certeza.

\say{Hi.}

Hi.

@MarkGiraffe hi

Cool, it works.

\language{jp}

say{Hi.}

こんにちは。

\language{en-us}

\language{es}

\say{Hi.}

Hola.

\language{en-us}

I guess it works.

@DavidCarlisle ooh

arstechnica.com/tech-policy/2021/03/…

> LOL Garamond sux, say federal judges

> "The court has determined that certain typefaces, such as Century and Times New Roman, are more legible than others, particularly Garamond, which appears smaller than the other two typefaces," the DC Circuit announced this week. The court, it said, wants to "discourage use of Garamond."

@egreg ^^ OH NO

Last sentence of the article: WILL GARAMOND TOO BECOME A VICTIM OF CANCEL CULTURE?

> a

>ha

???????????????????

@PauloCereda I tested this, it seems right

@DavidCarlisle accurate :)

@DavidCarlisle What have the Times New Roman ever done for us?

@yo' ooh

MAILBOX, MAILBOX!

I made a program that makes me use TeX like programming functions.

\say{Hello, world.}

Hello, world.

\input{n="Hello."}

\say{n}

Hello.

@MarkGiraffe How about LuaTeX?

Someone needs a lesson on negation.