Working on ... things

@JosephWright ooh things

@PauloCereda Skype

... working on bringing us closer to the big party?

@samcarter ...or to release 3.1 and ruining the party forever ;-P

@Rmano Oh, forever party? That's a good idea!

@samcarter Is there going to be a big party?

@FaheemMitha I certainly hope so :)

@samcarter Not today

@JosephWright oh no :)

@samcarter you will have to console yourself with today's two easybook updates

@DavidCarlisle There is also elsarticle -- are only packages starting with e allowed today?

@samcarter no they can start with any letter, so long as the third letter is s

@DavidCarlisle too bad <insert secret name> (cc @PauloCereda) is not yet on ctan, it could be updated today :)

@samcarter ooh

If one sets: TeX Live Utility -> Configure -> Change Paper Size... = A4 , would \documentclass{article} with no options be expected to produce a A4 paper? It does not work for me. Maybe it is controlled elsewhere?

I am using fully updated macTeX

@hpekristiansen you need [a4paper] option to change thetext block to match A4, the config setting should change the media size internally in the PDF

@DavidCarlisle: I do not understand the terminology in your answer. "thetext block" and "media size internally in the PDF". Can you explain a little more - thanks

@hpekristiansen The TeX Live option sets the size of the 'physical' PDF, the [a4paper] option in the source tells LaTeX to use the correct margins for A4 rather than letter paper

I think I have an answer on site but in shortm I have the texlive config set up for A4 so if I go pdflatex sample2e I get a PDF with textwidth and textheight set to match US Letter (as I did not use a4paper but if I use ppdfinfo sample2e.pdf it says Page size: 595.276 x 841.89 pts (A4) and a pdf viewer would open it in a window withA4 aspect ratio (and thus unplanned margins)

Is default for no option \documentclass{article} always letter. If I want a A4 PDF that is A4 dimensions on my computer and can be printed on A4 paper, do I need both the [a4paper] option AND the setting in TeX Live?

If you use graphicx the backend knows it is on pdftex so sets the page size to useletter to match the text block or conversely if you use [a4paper] and load graphicx the text block is set for A4 and the PDF media size is set to match

@hpekristiansen yes (although you only need the texlive setting for the default. If you use graphicx or geometry the page size is explicitly set in each document)

One day @JosephWright will fix the l3backend to have more sensible defaults (but compatibility here is tricky)

@DavidCarlisle Yes, was just about to comment to that effect

I will have to study your answers to understand - thanks. I always(except for rare lua) use TeXShop using pdflatex to compile.

@hpekristiansen ultimately the problem is that the original TeX/DVI model doesn't really have any concept of paper size or margins. It just controls typesetting a rectangular block of text. The page size and any margins are controlled by the size of the sheet of paper that you put in your printer. Then PDF specification came out with an intrinsic notion of the media size and we have been trying to make latex catch up ever since....

@hpekristiansen lualatex and pdflatex are the same here

Well I love LaTeX - and I love the modern pdf over olden ps, eps, dvitops, ghostview and what not, that I have happily forgotten how to use

@JosephWright if you are fixing these 2-D layout parameters, you may also need to fix it for the new printer in our house (not mine:-) that has a Z axis to worry about....

@UlrikeFischer I'm sure they would jam the print nozzle

@UlrikeFischer do German 3-D printers have the y and z axes swapped?

@DavidCarlisle -- Uh oh ... "ams" has "s" as the third letter ...

@UlrikeFischer seen this failure before? tex.stackexchange.com/a/630746/1090

@DavidCarlisle no. I wonder if it is related to the format or specific to the font, perhaps (probably) @MarcelKrüger knows something.

@UlrikeFischer I just opended the type1 font in fontforge, it gives no warnings but is almost empty, just has letters A-Z

@DavidCarlisle there is another one but without a full example tex.stackexchange.com/questions/615300/…

@DavidCarlisle in the second example it is dictsym, and that is a pfb.

@UlrikeFischer writefont.c: pdftex_fail("builtin glyph names is empty"); lets see how it gets there...

@DavidCarlisle The problem is that the font does not get subsetted, but the glyph names only get initialized in the subsetting code...

@MarcelKrüger hmm

@MarcelKrüger so I suppose alternative answer is "use luatex"

@DavidCarlisle you can do \pdfmapline{=pigpen <pigpen.pfa}.

@UlrikeFischer that would be better enabling the font without removing glyphtounicode for everything else. This could be in the default map, I guess? Unless we want to suggest a code change to pdftex not to error.

@DavidCarlisle Another option would be to add \pdfnobuiltintounicode\font after \pigpenfont (or in the .fd file I guess) to disable glyphtounicode only for this font (for which it does the wrong thing anyway)

@UlrikeFischer I don't think so, but especially it shouldn't affect reencoded fonts which should cover most otf fonts anyway.

@MarcelKrüger good, I thought so, we would have heard imho if they had problems.

@MarcelKrüger so you suggest we suggest this is fixed at the package level with \ifx \pdfnobuiltintounicode\undefined \else\pdfnobuiltintounicode\font\fi added after the font selector?

@DavidCarlisle I think it would be better to change the map file to subset the font (especially since I don't see any reason not to subset it in the first place). Then it still gets weird ToUnicode mappings, but weird copy and paste will happen for this font even without the cmap. I'm also thinking about suggesting a pdfTeX patch, but I'm a bit concerned about side effects.

@MarcelKrüger arguably the pigpen one really is A-Z isn't it? so should map to that. they are normal letters with unusual shapes

@DavidCarlisle Oh, I see. I just assumed that it's some kind of symbol font. Then having the mapping makes sense.

@MarcelKrüger apparently some secret code from freemasons (not exactly the most crypotgraphically difficult code to crack:-)

@DavidCarlisle We should request a CVE: The choice of glyphnames allows PDF viewers to crack this highly secure encryption via a well hidden technique known as "copy&paste" :P

@MarcelKrüger @DavidCarlisle dictsym is from Walter Schmidt, so I guess we can simply change the map and upload? For the other we will have to contact the author imho.

Where are the Englishmen to defend their spelling? tex.meta.stackexchange.com/posts/430/revisions :)

@samcarter Englishmen have better taste and don't use annoying pre-written comments

@DavidCarlisle Your habit of pineapple pizza eating makes me doubt the better taste :)

@JosephWright I have a problem with the xcolor patches in the pdfmanagement. the following errors for me with pdflatex. It still works in a current miktex and I think a change in __color_convert:nnnN is the problem:

\RequirePackage{pdfmanagement-testphase}
\DeclareDocumentMetadata{uncompress}
\documentclass{beamer}

\setbeamercolor*{testcolor}{use=structure,bg=structure.fg}

\begin{document}
\tracingmacros=1
\usebeamercolor{testcolor}

\show\stop
abc
\end{document}

@UlrikeFischer @MarcelKrüger was in charge of those changes :)

@UlrikeFischer I'm investigating

@samcarter Not a big fan of such comments either. But we are supposed to have a "don't change spellings" like this. But it's not worth getting into an edit war over.

@AlanMunn Yup, agreed

meeting in 10 mins?

Ohh, meetings.

@DavidCarlisle Have fun

@JosephWright you skipping this one? ah now there are three

@DavidCarlisle Have to work

@AlanMunn There is a very easy solution: you just have to convince the people on your continent to change their behaviour -- problem solved once and for all :)

@samcarter -- "NBL!" (Not bloody likely. I'm afraid that I, for one, am too set in my ways.)

@barbarabeeton If this is not an option, then @DavidCarlisle needs to use his linguistic skills to modify the behavior of his compatriots :)

Just prize the u key off the keyboard

@samcarter -- Nah, that won't work either. Their kind have been around longer than the colonists, and are set in their ways. Maybe we should all go back to Anglo-Saxon.

@DavidCarlisle -- Bt then I coldn't answer you with "Nts".

@AlanMunn -- Totally off-the-wall question. The Tongan place names currently in the news appear to have a phonetic structure similar to that of Hawai'ian. Do you know whether they're related?

@barbarabeeton In this case my next suggestion is to meet in the middle and use a letter which looks like half an u (the existence of such a letter would only be logical, after all double U exists as well :) )

@samcarter -- That might require a newly designed keyboard. Have to think about it.

@barbarabeeton :)

@barbarabeeton Up or down? Which one do you prefer? :)

@barbarabeeton Yes, they're all part of the Polynesian branch of the Malayo-Polynesian language family.

@samcarter I'm from the part of my continent that already does that. I just so happen to live in the unenlightened part that doesn't. :)

@mickep -- Personally, I don't really have a preference, although maybe "down" is a bit more precise. Some purists (?) might suggest that the e should really be upright, and that would actually be meaningful. I don't really care for the extra space following the superscript though.

@barbarabeeton I think I prefer the lower one, in particular when the e is upright (tested that). (I'm reading the old Swedish guide for setting math, and it suggests space on both sides when it comes to the exponential function, but not on other powers, like if the a would have been a^2.)

@mickep -- I understand the suggestion, but I'm thinking that the extra space after the exponent looks nearly as wide as a word space might be if a line is set tightly. Perhaps this is just the size of the type in the example, and the lack of context. But (I think) worth considering.

@barbarabeeton Thanks! The picture has to much spacing, for the simple reason that I forgot that TeX adds space by default in supersripts. This means that I added too much (again, according to Lansburgh). The book is from the 60s, so it is Monotype setting they had in mind.

@mickep -- Can you supply bibliographic information for the Swedish guide? (I've got a rather nice collection of references re math setting, and my library will be housed at the Museum of Printing in Haverhill, Massachusetts. Relevant additions are welcome.)

@barbarabeeton It is this one: antikvariat.net/en/…. It is a very nice book, over 450 pages. As far as I understand it was used as a guide for that publisher, and they published the math journal Acta by then. It is very detailed, and with good(or interesting) arguments for the rules.

@barbarabeeton I'll paste half a page (of the over 30 pages) on how and where to put spaces in math.

@mickep -- Thanks! I believe that Acta was one of the journals that DEK used as a model for TeX. The publication date (1961) definitely puts it in the Monotype era.

@barbarabeeton At the bottom you can read "Undantag" (Exceptions). :)

@mickep -- Yes, I see. One other thing I notice is the position of the primes. (They're a bit tight and high. I'll have to look at Chaundy and compare.) I'll really have to acquire a copy of this book -- thanks!

@barbarabeeton Nice! I guess you could buy the copy I linked to. Or search on the same site for a cheaper one. If they do not ship to US you can tell me and I can buy it and ship it to you.

@barbarabeeton This copy looks nice (2nd edition, not sure what that means, except that it is from 1964): bokborsen.se/view/Lansburgh-W-N/…

@mickep -- What I intend to do is ask the bookseller I deal with in the U.S. (Oak Knoll, "books about books") to see if they can get it. The "overseas" postal charges can be outrageous, and recently I've had a bad experience when sending a book to a friend in England -- the package arrived empty! But I'll remember your offer!

@barbarabeeton Oh, I hope you'll find it!

@JosephWright the problem is a global/local color setting. beamer sets colors in parts in groups but globally. And this doesn't work well if the l3color patches are there too:

\RequirePackage{pdfmanagement-testphase}
\DeclareDocumentMetadata{uncompress}
\documentclass{article}
\usepackage{xcolor}
\begin{document}
{\globalcolorstrue\colorlet{abc}{red}}

\colorlet{blub}{abc}
\end{document}

@UlrikeFischer Ah, that would do it

@UlrikeFischer I guess that needs fixing in beamer, really

@JosephWright I have some doubts that all beamer colors really worked correctly previously but what is the change that it now errors?

@UlrikeFischer That I'm not sure: I suppose we might need to trap and issue an error

@JosephWright well the question is if this global color stuff is needed or not. If yes one would need something similar in l3color.

@UlrikeFischer I've not seen a case where I think it's necessary

@JosephWright it is used much less often than I thought. I can find it only in beamer and there only 3 or 4 times.

@UlrikeFischer I will do a few experiments

@UlrikeFischer I think we are seeing the issue as we have a prop for l3color data: we test if it exists but not if particular entry is there, so when we are at the wrong group level, we get an empty string back unexpectedly

@JosephWright and one occurance in tikzcodeblocks.sty.

@UlrikeFischer Something like:

\makeatletter
\newcount\beamer@smuggle@color@cnt
\def\bearm@smuggle@color#1#2{%
  \global\advance\beamer@smuggle@color@cnt 1 %
  \edef\@tempa{beamer@smuggle@color@\number\beamer@smuggle@color@cnt}%
  \expandafter\xdef\csname\@tempa\endcsname
    {%
      \noexpand\colorlet{#1}{\noexpand#2}%
      \noexpand\ifnum\noexpand\currentgrouplevel=0 \noexpand\else
        \noexpand\aftergroup\expandafter\noexpand
          \csname\@tempa\endcsname
      \noexpand\fi
    }%
  \expandafter\aftergroup\csname\@tempa\endcsname

at the beamer end would work for this case, and I could extend. Of course, needs proper testing ...

@JosephWright yes it is actually independant from xcolor:

\documentclass{article}
\begin{document}
\ExplSyntaxOn
{\color_set_eq:nn{abc}{red}}
\color_set_eq:nn{blub}{abc}

\color_select:n{blub} xxx
\ExplSyntaxOff

\end{document}

@UlrikeFischer Cool: one for @MarcelKrüger then (actually, I may have a solution soon-ish)

\cs_gset_protected:Npn \__color_parse_set_eq:nNn #1#2#3
  {
    \prop_get:cnNTF
      { l__color_named_ #3 _prop } {#1}
      \l__color_value_tl
      { \tl_set:Nx #2 { {#1} { \l__color_value_tl } } }
      {
        \tl_set_eq:Nc \l__color_model_tl { l__color_named_ #3 _tl }
        \prop_get:cVNTF { l__color_named_ #3 _prop } \l__color_model_tl
          \l__color_value_tl
          {
             \__color_convert:nnN
               \l__color_model_tl {#1} \l__color_value_tl
          }
          { \ERROR }

@UlrikeFischer ^^^ Needs something more sensible, but the \ERROR is where we need to trap

\cs_gset_protected:Npn \__color_parse_set_eq:nNn #1#2#3
  {
    \prop_get:cnNTF
      { l__color_named_ #3 _prop } {#1}
      \l__color_value_tl
      { \tl_set:Nx #2 { {#1} { \l__color_value_tl } } }
      {
        \tl_set_eq:Nc \l__color_model_tl { l__color_named_ #3 _tl }
        \prop_get:cVNTF { l__color_named_ #3 _prop } \l__color_model_tl
          \l__color_value_tl
          {
             \__color_convert:nnN
               \l__color_model_tl {#1} \l__color_value_tl
          }
          { \msg_error:nnn { color } { unknown-color } {#3} }

@UlrikeFischer ^^^ Probably this

@JosephWright looks right, and together with your beamer code my original example compiles now.