« first day (3823 days earlier)      last day (28 days later) » 

6:43 AM
@JosephWright sorry I didn't really pay attention when compiling, I only made a fast check if it compiles at all.
7:15 AM
@JosephWright interesting... lvjr.bitbucket.io/tabularray.html
@DavidCarlisle oh yes.
8:13 AM
Hi everybody, I have a little problem with a left square bracket.
Is there anybody that could help me, please? It's a really trivial question to be honest, but it's the first time that I use this syntax.
Thanks in advance!
@LukeTheWolf no one can help if you don't say what the problem is (if it's more than a line or two of code, usually best to ask as a question on site rather than here, but try here first if you want...)
the answer is probably to put \relax in front of it.
@DavidCarlisle I'm really sorry, I already had the code ready but my intention was to see if there was anyone who wanted or had time to help me even with something so simple!
\text{alfabeto} & \x = \{0,1\}\\
\text{densità discreta} & p_X(1) = p, p_X(0) = 1-p\\
8:31 AM
@LukeTheWolf \left[ and \right.
@UlrikeFischer ah I made the wrong guess :-)
I've initially done that, but my compiler says "Missing delimiter (. inserted).
<to be read again>", that's why I'm asking you this :D
@LukeTheWolf remark the period after the \right. <--- that is part of the code.
@UlrikeFischer Oh, I didn't see the actual function of that full stop, TY
8:46 AM
@JosephWright I'm trying to extend a key in tagpdf to make it pdf 2.0 compatible. I wanted to handle the old syntax as "unknown" keys but it fails as there are slashes involved. Is there any chance to get the following working without errors? (If not I will create a new key)


\keys_define:nn { main }
  a .code:n       = {known},
  unknown .code:n = {unknown}

\keys_set:nn{main}{x/y}  %fails


Well, a name change is needed. Hmmm... clabel distance? ctikz label distance seems too long, no?
@JosephWright I think I can use keys_set_known in my setup, will try now.
@Rmano blame the user
@Rmano which name?
8:53 AM
@DavidCarlisle that's an option (it's true that I say that compatibility with internal circuits is not guaranteed), but using label distancehas been a bit bold on my side
@UlrikeFischer I used label distance...
@Rmano yes, I just saw your link to the issue. But don't you have your own path?
Maybe I have to bit the bullet and use ctikz label distance, so I can blame others if it clashes...
@UlrikeFischer yes, but for commodity, I move it also on the top level, which is the error really. I will remove that!
9:29 AM
A good day to you all, friends
@DavidCarlisle In MathML Core (Appendix B.4) is a table of "Combining Character Equivalences" but I couldn't find any reference to it. Is there any rule about which version should be used for accents etc?
@JosephWright could we change the value of \pdf_version_major: to -2 in dvips if version 2.0 was set? Then I could offer also for dvips the option to get pdf2.0 code. It wouldn't be fool proof, but the users had a chance at least.
@UlrikeFischer I guess
10:02 AM
@MarcelKrüger something we probably need to refine (the working group is just restarting now) the original mathml1/2/3 idea was that as far as possible you shouldn't use combining characters and so over would be <mover><mi>x</mi><mo>.</mo></mover> but in practice users use them anyway so that is the start of a table that will in the end say what a mathml reader is supposed to do if it gets accented characters or combining accents in the stream.
@UlrikeFischer -2 ?
@DavidCarlisle well currently \pdf_version_gset:n sets major/minor to -1 to indicate that it actually can't set the version in dvips. But I would like to have a least an idea which version the user wanted.
@UlrikeFischer ah OK
@DavidCarlisle Thanks. So mover would use neither one and just use the simple .? That seems odd from a semantic point of view and possibly inconsistent since it wouldn't work for some other accents. But I'll wait for a more final spec.
@DavidCarlisle but perhaps it would be better if the code wouldn't simply set major/minor to the requested values if \pdf_version_gset:n is used (perhaps with a warning that it doesn't really set the version and that the command line option is needed). It would make at least my code easier.
@MarcelKrüger readers have been very flexible in what characters they accept eg an overbar taking a macron character or - or _ or ... Some accents are only available as combining characters and combining accents are a bit problematic in xml as <mo>combining accent</mo> could potentially end up with it being normalised to the accent applied to > this doesn't hit in practice as the only such mapping is the negation slash which normalises to not greater-than so making your xml non-wellformed
... but that's by luc rkather than by design and theer was a feeling that we should specify what systems are supposed to do not just leave it to teh heuristics in each reader, but the current table is just a sketch
@MarcelKrüger you are writing a mathml reader?
10:18 AM
@DavidCarlisle No, I'm working on the other side. I revived an older project of mine to write a LuaTeX mlist to MathML converter.
@MarcelKrüger ah. A good general policy would be to favour the characters that have entity names (not necessarily to use the names, but just because those are the ones we decided to favour last time we thought about this) you can pull those out of unicode.xml or from the derived json or html files from mathml-refresh.github.io/xml-entities eg w3.org/2003/entities/2007/w3centities-f.ent
@MarcelKrüger it would be nice if we could process some math with luatex and get the mathml.
1 hour later…
11:39 AM
@UlrikeFischer That's the plan. It's mostly an early draft currently, but feel free to play around with it (git.math.hamburg/zauguin/mathml)
1 hour later…
12:56 PM
@MarcelKrüger it looks quite nice. How much does it depend on unicode-math? (I naturally see that symbols like \int are output as R without it, but the structure seems fine nevertheless)
@UlrikeFischer It currently just assumes that the codepoints are unicode values, but it could be extended to remap other encoding too. Structurally it shouldn't make a difference.
1:18 PM
There is a bug in my code - and I don't think that's a problem :)
user image
@samcarter_looks_forward_TUG'21 Isn't that what all code changes do?
@PhelypeOleinik Yes - it is time to be honest about it :)
@samcarter_looks_forward_TUG'21 :-)
@samcarter_looks_forward_TUG'21 Nice bug, by the way :)
@PhelypeOleinik Thanks! No more hiding behind calling things features :)
1:33 PM
I discovered that the combination \usepackage[symbol]{footmisc}
\usepackage{newtxtext} messes up the vertical placement of footnotemarks. One can direct newtxtext to [defaultsups] to avoid the problem by not changing the script fonts. Is this worth letting a package maintainer know about?
@StevenB.Segletes footmisc is Frank these days you could add a github issue blaming him at github.com/FrankMittelbach/fmitex/issues
@DavidCarlisle the problem is more newtxtext. It redefines \@makefnmark. One can avoid the problem also by loading footmisc later. But we really need more hooks there so that packages stop to redefine all this stuff and can simply add their formatting to make the marks blue or with a pattern or whatever deko they want.
@UlrikeFischer sounds like it should be best reported to Frank and he can pass on to newtxtext if needed
1:50 PM
@UlrikeFischer, @DavidCarlisle I see we all agree on the PDF version stuff
@DavidCarlisle Thanks!
@JosephWright yes. Should we set it to a sensible start value like 1.5 or leave it at -1.-1?
@UlrikeFischer I think unless it's been set we shouldn't have a positive value
Hi everybody, I'm trying to insert a \twocolumn text inside my article.
The problem is that the part starting with the twocolumn command and ending with the onecolumn command is inserted in a new page, but I would like everything to be inserted on a single page.
Below is my code

\documentclass[a4paper, 12pt, italian]{article}
\geometry{legalpaper, margin=1in}
@JosephWright fine with me.
1:56 PM
@LukeTheWolf take a look at the multicol package.
@Skillmon I will, thanks a lot
2:24 PM
@LukeTheWolf you can out the spanning part inside \twocolumn[.... here]
For some reason one of my answers is marked as deleted (the question got deleted by Community because it was closed; but I didn't delete my answer and I have no idea why it is marked as deleted).
Is that expected behaviour for a closed question? I thought answers aren't auto-deleted.
@DavidCarlisle Forgive my ignorance (this is my first year using LaTeX), but I did not understand what you mean
@Skillmon There are some rules about votes, views, etc.: link?
2:36 PM
@DavidCarlisle @UlrikeFischer on the footnote hook problem, maybe I'm overthinking this. Is it sufficient to just use \pretocmd{\footnote}... to do what I want? It seems to work. Am I missing any potential problems with this?
@JosephWright not that it is a great loss... I'm just surprised by this.
@AlanMunn you are outside a group then aren't you?
@Skillmon I think it's because the question is a dupe with negative score and was then deleted by the OP
@JosephWright but do answers get marked as deleted in that case as well? I thought they are then just hidden as part of the deleted thread, not marked individually.
@UlrikeFischer Yes, but I can set my boolean in the code I add.
2:38 PM
@AlanMunn yes, but can you reset it after the footnote?
@Skillmon Well yes I see what you mean, but all answers to deleted questions are themselves marked as deleted
@UlrikeFischer Yes, because I can use the after key in enumitem to do the restore if the fn boolean is true, and then set it to false again.
@AlanMunn and what if you have two lists in the footnote?
@UlrikeFischer enumitem allows to set options individually based on nesting levels, iirc.
@UlrikeFischer Oops :) Yes, that's a problem.
2:42 PM
@Skillmon I meant two lists one after the other.
@UlrikeFischer ahh, now that makes a lot more sense :) Ignore me.
3:00 PM
It seems that every year I ask this: how can I do a final update on TL2020 now that the 2021 is released?
@LukeTheWolf \twocolumn[\section{zzz} text here spans both columns] two col text starts
@DavidCarlisle Ah, ops. I got it now ... Sorry for that
By the way, I solved this problem using multicol package as suggested before
@LukeTheWolf simpler is to use report class and \chapter which automatically spans both columns
@LukeTheWolf yes OK but multicol is different kind of two-column
@DavidCarlisle I just needed to do this kind of thing, so multicol worked perfectly (but maybe it isn't the right command) : ibb.co/d4nRVZB
3:38 PM
May 9 '19 at 15:52, by Paulo Cereda
@AlanMunn Let it go let it go
@samcarter_looks_forward_TUG'21 Yes, sometimes the quality of helpful advice here is amazing.
@AlanMunn :) sorry, I could not resist!
@LukeTheWolf really depends on what you want. If you need contents on the same page after the two columns you can't really use \twocolumn, that one only supports spanning content at the top, if you want to use \twocolumn for a segment of text, only need spanning material before the two columns and no content after them on the same page that would be correct.
@samcarter_looks_forward_TUG'21 But somehow I did find a solution, because the frozen repo from 2018 was still listed in my TeX Live Utility repo list so I just changed it to 2020 and it worked.
3:43 PM
@AlanMunn quack <3
As you mentioned, I actually need contents both before and after the two columns, so the easiest way I could solve this thing (as you said before) is to use multicol.
Thank you for your explanation!
@samcarter_looks_forward_TUG'21 Yes, that's more or less what I have (but tug.org/historic/systems/texlive ) instead of the utah address.
@DavidCarlisle the change we made for github.com/latex3/hyperref/issues/167, introduce also break points after whatsits like e.g. \addcontentsline or color. Would it make sense to use \nobreak also if the node type is 9? Or will that break something else?



\color{red}%breakpoint ...

user image
siunitx 3, chapter 3
(WhatsApp duck follows Joseph)
3:58 PM
@PauloCereda Quack
@UlrikeFischer I'd have to remind myself of details but it might be better to list the ones where we don't add nobreak rather than the ones where we do...
@UlrikeFischer Back to hooks. Your code didn't work for me (produces blub inside the footnote).
@AlanMunn oh no it can't work for you. That's secret code ;-)
@DavidCarlisle the older code had \nobreak always, so probably it is really better to add the breaks one by one. The minipage case is 1 isn't it?
4:20 PM
@UlrikeFischer sorry can't look now
@DavidCarlisle no problem, I already checked. I will change the code so it only breaks for 1, and if we find other cases one can enlarge the set later.
5:05 PM
@CarLaTeX @samcarter_looks_forward_TUG'21 vvvv
very bad quality because I was cycling and with my lousy cellphone (never a camera when you need it) --- but that lassie had 10 little ones with her!
@Rmano 10 dinners in a few weeks
@samcarter_looks_forward_TUG'21 nice avatar
@DavidCarlisle well, appetizers probably ;-P
@Rmano WOW!
5:46 PM
@UlrikeFischer Ah, ok. Unlike @PauloCereda you're good at keeping secrets. :)
6:03 PM
@DavidCarlisle So slow :)
6:37 PM
@AlanMunn ooh a secret
@AlanMunn ;-). But it is not really a secret, you could find it at the github.
7:21 PM
What new thing from the TeX/LaTeX world should I know today?
@LaTeXereXeTaL you should know how many times to run latex on this document to get a correct page reference.





more text\\
more text

more text\\
more text\\
more text

more text\\
more text\\
more text\\
more text\\
more text\\
more text

more text\\
more text

There is some text, in
Section~\ref{z} on page~\pageref{z}.
That has a, b, c.

A, B, C.
@DavidCarlisle I run it until I see the PDF file in my previewer!
@LaTeXereXeTaL not a good plan
@DavidCarlisle Oh I figured that. Otherwise I have no answer.
@DavidCarlisle what is the problem?
7:32 PM
@UlrikeFischer try it:-)
@DavidCarlisle I did, what I meant is, what is the source of the problem? Why doesn't it stop?
@DavidCarlisle oh I see, a different line break.
@UlrikeFischer normal example uses roman page numbers and page iv being wider than page v but this is a bit less obvious and uses fewer pages as 2 is wider than 3
@DavidCarlisle I'm up to about six.
@LaTeXereXeTaL hint: it is an infinite loop
@DavidCarlisle I guessed that after the first two runs.
@DavidCarlisle Putting \\ after A, B, C. seems to fix it, but I don't understand why.
7:43 PM
@LaTeXereXeTaL putting \\ at the end of a paragraph is evil though.
@DavidCarlisle I know, but it's done elsewhere in the example.
@LaTeXereXeTaL that "fixes it" as there is not room to have a two line paragraph
@LaTeXereXeTaL no, there isn't otherwise you would get a badness 10000 warning when running the example.
@LaTeXereXeTaL there are \\ but not at the end of a paragraph
@DavidCarlisle Lines then.
The blank lines end/start paragraphs.
@LaTeXereXeTaL that is completely different a forced line break is the intended usage of \\ but \\ at the end of a paragraph forces an empty spurious line and a badness 10000 warnings, and as you demonstrate messes up latex's count of lines for widow control after section headings.
@DavidCarlisle I just generated the badness 10000.
7:53 PM
@LaTeXereXeTaL so by definition that can't be a fix:-) (although (\\zz would also work without generating a warning)
@DavidCarlisle I did it by putting \\ on one of the blank lines.
@DavidCarlisle Got it before and didn't see it then.
@LaTeXereXeTaL yes as you say, it also made latex think there was a two line paragraph after the last section so there was not room for it at the bottom of the page, so it moved it to the next page (so avoiding the pagref loop) but there is not really a two line paragraph it is just one line A B C. and one spurious underful line generating the warning that is forced by the mis-placed \\
@JosephWright ^^ is that user error or a bug? It is from

@DavidCarlisle How does one fix something like this?
@LaTeXereXeTaL it only occurs if you have a paragraph whose line breaking changes if you change a 2 to a 3 which is narrower than the 2, and if that change on line breaking changes the page breaking and the page on which the referenced section occurs changes from page 3 to page 2. So the hard part is making it occur not fixing it.
@LaTeXereXeTaL to put it another way if you change the text or spacing in more or less any way at all the problem will go. If you do not change the text or spacing the problem is not fixable as the document constraints can not be resolved.
8:11 PM
@DavidCarlisle You mean LITERALLY the difference in width between a 2 and a 3 causes this?
@LaTeXereXeTaL yes and in particular the fact that in this font the 2 is wider than the 3
@LaTeXereXeTaL well look at the example, you can see how the lines change.
@UlrikeFischer in your luaotfload testing you haven't seen a font with 1 wider than 2 have you? that would allow the example to have less spurious text at the start...
@DavidCarlisle hm, no. But can't you simply start the page numbering with 2?
@UlrikeFischer the aim of the game is to make it look accidental
8:23 PM
@UlrikeFischer I can now, but I would never have thought to look for that before. I would have assumed such problems could not happen.
@LaTeXereXeTaL you said you wanted to learn something about latex today:-)
@DavidCarlisle And I certainly did!
I'm keeping this example.
Could this not be considered a bug?
@DavidCarlisle Just use \pagenumbering{Alph}, CMR has A wider than B.
@LaTeXereXeTaL no, no more than \def\z{\z}\z looping is a bug.
@MarcelKrüger I know, but who numbers pages in capital letters?, that's just cheating.
@DavidCarlisle Interesting!
@DavidCarlisle Commenting out the \renewcommand line prevents the problem, so it depends on typeface?
8:28 PM
@LaTeXereXeTaL a version with roman page numbering is on site it is much easier to make that example as iv is a lot wider than v:
A: Document requiring infinitely many compiler passes?

David CarlisleHaving an explicit \ifthenelse test makes the document look like the looping is intentional. It is possible for the cross referencing not to converge even if there is no explicit conditional switching within the file. This document for example. \documentclass{article} \pagenumbering{Roman} \beg...

@LaTeXereXeTaL yes most standard tex typefaces use tabular digits that are all the same width so you can line them up in columns and many fonts with proportional digits have 3 wider than 2, so you have to work to find a font with 2 wider than 3
@DavidCarlisle I bookmarked that Q and A. Well, I certainly did learn something.
9:10 PM
Hi everyone! What's a good package to generate random numbers outside LuaTeX? I want to emulate the behavior of math.random and math.randomseed :)
@JairoA.delRio \fp_eval:n{rand()}
@PhelypeOleinik Is this built in LaTeX/expl3?
@JairoA.delRio yes.
@JairoA.delRio Yes, part of l3fp. Running \ExplSyntaxOn \typeout{\fp_eval:n{rand()}} \stop should work, if your LaTeX is at least from 2020
@JairoA.delRio Or \RequirePackage{xfp} \typeout{\fpeval{rand()}} \stop in older LaTeX
@PhelypeOleinik Cool. Thanks!
10:18 PM
@UlrikeFischer User error
@JosephWright but it passed previously (apart from some overfull box messages).
@UlrikeFischer Well that's the same: it was badly formatted and it still is: there's just not enough space reserved
@UlrikeFischer It doesn't 'pass', the alignment is out
@JosephWright yes I know, and I actually wanted to solve the overfull boxes for some time (the numbers got larger over time ...), and the now quite dramatic overlapped forced me to do it ;-)
10:48 PM
@PhelypeOleinik Just another question, it allowed. Is there a macro for md5 hashes? In interface3 only files, not strings, are considered. I know how to define my own, but I'd rather not reinvent it. I mean something like pdfTeX's \pdfmdfivesum
10:59 PM
@JairoA.delRio expl3 has only a file interface for \pdfmdfivesum, so you need to use the primitives for a generic string
@JairoA.delRio If you want cross-engine you can abuse \tex_mdfivesum:D, but it's not really recommended to use a :D function
@PhelypeOleinik Yep, that's what I need. Off-topic: I find those happy faces at the end of csnames menacing
@PhelypeOleinik Thank you very much
@JairoA.delRio I find them so joyful that it's hard to resist using them where I shouldn't ;-)
@JairoA.delRio If \tex_mdfivesum:D does what you want, then maybe I can convince you to use \pdf@mdfivesum from the pdftexcmds package instead. It does more or less the same thing, but it's a documented interface
11:47 PM
@FaheemMitha -- The principle is the same with metal type. TeX doesn't matter, actually. Consider a line 20cm wide. Fill a line that width with type, say all "x"s Then consider a line 30cm wide. Fill that line with type, also all "x"s. Even without counting the number of "x"s, it should be obvious that there are more on the 30cm line than on the 20cm line. That's true for TeX pages -- narrower lines, less type, so more lines needed for a paragraph, hence more pages for the same text.
11:57 PM
@AlanMunn and @DavidCarlisle -- Would it be possible to clone the footnote mechanism under another name, say AMfootnote and include the special code there? With another name, nobody is likely to redefine it. It's just another \botinsert I think (she says without checking).
@barbarabeeton usually with these questions you need to patch the standard one, ie clone it under the same name, ie have the user use commands from standard latex or footmisc or reledmac or whatever but still patch in your special footnote code. Just having a private macro takes all the fun out of it

« first day (3823 days earlier)      last day (28 days later) »