TeX, LaTeX and Friends

daleif

4:58 AM

@barbarabeeton are you just adding qedhere or tag{qedhere} to empheq or are you talking the normal case without empheq in play?

barbara beeton

5:43 AM

@daleif -- I tried several approaches, with and without the \tag \qedhere by itself places the tombstone on the baseline. But with empheq, the \tag approach fails with an error. I didn't find the example I remember that embedded a picture in an equation, and used a different mechanism to force the tombstone to the bottom. I can send you the file I used to test, and you can try it out for yourself. I'm going to sleep now; I'll send it in the morning.

mickep

6:24 AM

@daleif One can also argue that the box sitting on the baseline looks more correct. The second one looks a bit arbitrarily placed to me.

cfr

every time I think I'm done with ifthen, I find somewhere I can't figure out a substitute. what is the safest thing to substitute?

Joseph Wright

7:23 AM

@cfr Well expl3 can do everything that ifthen can, or if you want classical catcodes then etoolbox covers a lot

cfr

7:48 AM

@JosephWright I wanted to replace \ifthenelse{\equal{}{}}{}{} but I can't figure out an equivalent.

(I meant to write \ifthenelse rather than just ifthen above.)

from fancyhdr.pdf:

daleif

@mickep I don't agree when the equation has a larger depth, then it looks awkward

cfr

\newcommand{\mymarks}{
\ifthenelse{\equal{\leftmark}{\rightmark}}
{\rightmark} % if equal
{\rightmark--\leftmark}}

Joseph Wright

@cfr Well that's \tl_if_eq:NNTF, although most of the time string comparisons are more useful

daleif

@barbarabeeton I'm not sure empheq has tag support. Remember that internally it uses aligned and friends and then build eq no's separately.

mickep

@daleif OK, we then disagree. Let me just add: Placing the tombstone all the way down is to me a bit like placing an equation label in the middle between the two lines of a two-line formula.

daleif

8:05 AM

@mickep Which a lot of people do.

mickep

@daleif Yes, and here a lot of people walk against red light... :)

David Carlisle

@daleif the second one looks very low to me

mickep

@DavidCarlisle Oh no, shall we be on the same side, then there is no way to tease you...

David Carlisle

@mickep I'm sure you will find a way

daleif

8:56 AM

To me the first look horrible, and placing it like proof from amsthm seems such a waste of space. I'd like to give the users the ability to choose.

Ulrike Fischer

9:09 AM

@daleif well you have only a sum in your example where it is difficult to see the baseline, if you put some text after it, it looks better with the box on the line. But if one additionally put a box around the equation then the lower box looks better. So imho it depends on the content of the equation and there should be a choice.

David Carlisle

9:37 AM

@daleif as Ulrike says in any real case you have stuff after the summation and putting the box on the last baseline rather than on a line on its own seems more normal. Your second version has the box floating (or perhaps better sinking) weirdly disjointed from the expression

daleif

2:15 PM

@UlrikeFischer one could imagine qedhere=baseline and qedhere=bottom options.

David Carlisle

2:43 PM

@daleif or an optional shift parameter (like amsmath \raisetag)

daleif

3:07 PM

@DavidCarlisle which annoyingly only works if amsmath have moved the tag. Wiuld be nice if that limitation was removed. I do not remember why we wanted to manually move it (perhaps it was a \tag*{\qedhere} situation and \raisetag did nothing at all.

Skillmon

In German a Reisetag would be a day of travelling.

Pablo González L

3:23 PM

@DavidCarlisle @UlrikeFischer \l_tmpa_tl :D

\DocumentMetadata{lang=en-US, pdfversion=2.0, testphase={phase-III,math,table,title,firstaid}, uncompress}
\documentclass{article}
\ExplSyntaxOn
\tl_show:N \l_tmpa_tl % > \l_tmpa_tl={Span}{pdf}.
\ExplSyntaxOff
\begin{document}
Duck :D
\end{document}

David Carlisle

@Skillmon one day we will teach you about spaces

@PabloGonzálezL yes, what is the question?

Pablo González L

@DavidCarlisle Well it only took half an hour with a small test using tl_put without understanding what was going on...l_tmpa_tl is always supposed to be clean.

David Carlisle

@daleif it works if amsmath has positioned it (not for example for \[ or \begin{equation}

@PabloGonzálezL no it's always supposed to be unreliable and possibly used by some other package

@PabloGonzálezL there is no reason to use the scratch macros in your own code unless you are trying to stay with emtex's upper limit of csnames

Pablo González L

@DavidCarlisle I use it in small tests and then replace it with the relevant variables.

David Carlisle

@PabloGonzálezL ok but you can't assume any initial value

Pablo González L

3:39 PM

@DavidCarlisle Yes, but here it's cheating ...I didn't expect tagPDF to have it busy :(

David Carlisle

@PabloGonzálezL that is user error on your part then, even though, being tagpdf, you may be tempted to blame @UlrikeFischer

Skillmon

@DavidCarlisle Not to confuse with a Reistag, which would be a day on which we eat rice.

David Carlisle

@Skillmon what about days where you eat rice while travelling, I hope you dont tell me you give up and start adding spaces?

Skillmon

@DavidCarlisle that would be a Reisereistag, not to confuse with a Reisreisetag which would be a day on which we travel for rice.

David Carlisle

@Skillmon all sounds reasonable, and not a single spurious \relax added at the start.

Pablo González L

3:43 PM

@DavidCarlisle Me parece una excelente idea :)

Skillmon

Oh, and also you must not confuse a Reisetag with a Tagesreise, the former is one day of a travel that might be longer, the latter is a travel that takes a day (or less).

@DavidCarlisle well, we didn't need to collect any rice since we brought it with us, so there was no spurious \relax to be grabbed. (?)

David Carlisle

@Skillmon fortunately you didn't need a table

Ulrike Fischer

@DavidCarlisle hm, must be a leftover from some test. I need to check that.

David Carlisle

@UlrikeFischer yes but even then it's a "be nice to @PabloGonzálezL" not "bug fix"

Ulrike Fischer

@DavidCarlisle well we claim that the kernel doesn't use them, and tagpdf is starting to be "kernel".

David Carlisle

3:57 PM

@UlrikeFischer yes but that's in interface3 so means l3kernel anything that happens after that is fair game

cfr

@JosephWright but it doesn't give me the same result.

 }
 \cs_new_nopar:Nn \__cfrlm_doc_lreq:
 {
-    \ifthenelse { \equal { \firstleftmark } { \firstrightmark } } {} { \firstrightmark }
+  \tl_if_eq:NNTF \firstleftmark \firstrightmark {} {\firstrightmark }
+    %^^A \ifthenelse { \equal { \firstleftmark } { \firstrightmark } } {} { \firstrightmark }
 }
 \fancyhf[rh]
 {

David Carlisle

@cfr you want ee or VV not NN to expand the marks (probably)

@cfr ifthenelse does a test after a \protected@edef so you would need to locally set \protect to \noexpand (or not support it) then use e expansion to get the same test

cfr

@DavidCarlisle with ee I definitely get an error. I'll try VV

David Carlisle

@cfr a MWE would help but ee should be OK as long as \let\protect\noexpand

cfr

@DavidCarlisle gives a different wrong result. I'll try to make a smaller example.

Ulrike Fischer

4:22 PM

@cfr you can naturally do a \protected@edef first, but why is that needed? if they are different a simple \tl_if_eq:NNTF should be ok?

cfr

5:03 PM

@UlrikeFischer it gives the wrong result.

Jasper Habicht

@Skillmon Don't forget to bring you box of Reisreisetagesreis (which you can buy at a specialised Reisreisetagesreisgeschäft hopefully for a reasonable Reisreisetagesreispreis)!

mickep

@JasperHabicht Oh, German humor!

Jasper Habicht

@mickep =) Right, I hear nobody laughing, so it was an excellent German joke.

mickep

@JasperHabicht Oh, another one.

Jasper Habicht

5:44 PM

@mickep :o)

cfr

% \iffalse
%<*driver>
\documentclass[10pt,british]{ltxdoc}
\EnableCrossrefs
\CodelineIndex
\RecordChanges
\usepackage{babel}
\usepackage{extramarks,fancyhdr}
\usepackage{ifthen}
\usepackage[a4paper,headheight=14pt,marginparwidth=45mm,hmarginratio=4:1,vscale=.8,hscale=.7,verbose]{geometry}
\usepackage{kantlipsum}
\ExplSyntaxOn
\pagestyle{fancy}
\fancyhf{}
\renewcommand \sectionmark[1]{\markboth{\thesection{} ~ #1}{}}
\renewcommand \subsectionmark[1]{\markright{\arabic{subsection} ~ #1}}
\fancyhf[lh] {

Skillmon

@JasperHabicht I heard they have a special offer on Thursdays, it's called the Reisreisetagesreistagespreis.

2

cfr

@DavidCarlisle ^^

in this version the header at the top of page 3 is wrong. in other variations, the header in the index (the last page in this case) is wrong. or I can generate errors instead of erroneous output.

@Skillmon why doesn't de hold the records for longest place names? it seems an odd achievement to overlook given the linguistic features available.

@UlrikeFischer natural to you ....

Jasper Habicht

@Skillmon Nice!

Skillmon

@cfr Because there's no reason to be this precise when it comes to city names. Most of them are historic names, and the people in the past were not likely to say a name for 2 minutes straight. There are composite words as names for places, but those usually are only made using two nouns or maybe three (e.g. Schweinfurt means "pig's ford", or names ending in "-hausen" which basically means place where someone lived). Some places have long names, but still in the reasonable range.

Jasper Habicht

5:58 PM

@cfr Well, of course we Germans could easily have the longest place names, but we are also known to be efficient, and this would just not be efficient ... no, we are just lazy and try to avoid too long place names I think =)

2

It seems that Hellschen-Heringsand-Unterschaar is the municipality with the longest name in Germany. And one of only two municipalities with two hyphens.

cfr

@JasperHabicht which really doesn't stand a chance in international competition.

Jasper Habicht

6:14 PM

@cfr Certainly not! It is truly embarassingly short!

cfr

@Skillmon obviously you have to name very small places with very long names. that way, you get your records while inconveniencing as few people as possible.

Skillmon

@cfr Well I live in a town with less than 2000 residents with the lovely name Waldeck (forst corner). I don't want it to be renamed.

@JasperHabicht I always liked the name Weil der Stadt (birthplace of Johannes Kepler) roughly translated as "Well, the city" but in German the word "Weil" is nowadays understood as "because", so it's "because the city".

cfr

@Skillmon well, no, I wouldn't want to live with such a name, though you can always shorten it for official purposes and keep the long version for the tourists :-).

Jasper Habicht

@Skillmon Yes, indeed a strange name

@cfr But in Germany, you are doomed anyways, because it is very likely that you visited some school named Annette-von-Droste-Hülshoff-Gesamtschule and you need to state that in every CV throughout your lifetime ...

cfr

Jasper Habicht

6:25 PM

@cfr Such signs normally have only two poles, I assume?

cfr

@JasperHabicht :-)

@JasperHabicht they're often attached to the wall, but otherwise, yes. they also don't typically include pronunciation hints for eng speakers. (regardless of how much such hints may be needed.)

Jasper Habicht

@cfr Well, with 4 L in a row, a pronunciation hint is surely a good idea ...

cfr

weird the signs below are monoglot english.

Jasper Habicht

@cfr Maybe they did not find paper that is large enough

cfr

@JasperHabicht that's only 2 letters. 'll` is a single letter.

Jasper Habicht

6:31 PM

@cfr I see

cfr

@JasperHabicht :-)

@JasperHabicht we have lots of them. I guess like 'ss'? but there are no single glyphs to represent them. so we have fewer glyphs but more letters in the alphabet. makes for efficiency in font design.

David Carlisle

@cfr you also have the cunning habbit of changing the spelling of the town depending on the context, just to make sure tourists stay lost

cfr

@JasperHabicht 'll' is the most notorious letter in Welsh ;).

Skillmon

@cfr hm, in Germany no one (I know) would call double consonants a single letter.

(at least not when speaking about the German language and its orthography)

Jasper Habicht

@cfr Yeah, well, sometimes ch is also considered one letter, or even sch, but they are also not single-glyph letters. But in Fraktur, ch and ck where represented as obligatory ligatures

Right, nowadays we would not say that ch is just one letter. It is a bit different

cfr

6:37 PM

@DavidCarlisle naturally. 'to Llanfairpwll PG' -> 'i Lanfairpwll PG'. but Cardiff if more interesting. 'i Gaerdydd', 'yng Nghaerdydd', 'Casnewydd a Chaerdydd'.

@JasperHabicht 'ch' is another single letter. so 'ch' is sorted after all the 'c's and has its own section in a dictionary.

Jasper Habicht

@cfr In German, you would not do that. China comes before Cider.

cfr

I don't think we have threes, though. but 'c' mutates to 'ngh' in some contexts. but only 'ng' is considered a single letter. (and it is a weird single letter which never starts an unmutated word.)

@JasperHabicht I didn't think so. but, yes, 'chi' comes after 'cyn', 'llawer' comes after 'lolipop' and so on. pre e-dictionaries, it made look-in-the-dictionary ... interesting.

@JasperHabicht interesting. that makes sense of why those letters tend to get ligatures, maybe. (or maybe not.)

Rmano

@cfr that happened also in Spain's Spanish, where ch and ll where considered single letter. A relatively recent (no idea when) normalization removed that, though. I've seen (at my in-laws home) some dictionary with the traditional order. Messy, especially when the thing happened midword ;-)

Jasper Habicht

@cfr In German, there are two ways to sort ä, ö and ü: ä = a or ä = ae, which is also a bit strange sometimes as you never really know which sorting rules are applicable

cfr

@DavidCarlisle we also change the name of England. and use names involving the 'll' for both you country and capital. whereas we're very nice to the scots: easy name with no mutations ever. (but admittedly it's often 'the Scotland' and the word for 'the' changes a little tiny bit now and again.)

Skillmon

6:52 PM

@cfr well, German also has a ss ligature which actually sticked, it looks like this: ß

@JasperHabicht I'd personally go the ae route though. I think it's more clear.

cfr

@JasperHabicht aaah. we're consistent at least. that seems like a way to confuse yourselves as well as @DavidCarlisle.

Skillmon

@cfr So you use Edinburgh for instance? Or only the name of the country?

Rmano

@Skillmon hmmm... A phonetically ordered dictionary for English would be... Interesting. 😉

Skillmon

@Rmano bear and beard are far away from each other! :D

Or lead and lead

cfr

@Skillmon well, no. and we don't use 'Scotland' but 'yr Alban'.

Skillmon

6:55 PM

@cfr ooh, so Scotland is basically the same as Albania?

cfr

@Skillmon that's what I meant earlier, only I don't know how to type it.

Rmano

@Skillmon 😂

Skillmon

@cfr well, it's right on my keyboard, so just come over and hit on that one.

cfr

@Skillmon apparently 'Albania' is 'Albania'. who would've guessed?

Skillmon

@cfr please pretty please tell me that the ending "-ia" is a belittlement or something in Welsh!

cfr

6:57 PM

@Skillmon :-) I can't type pound signs either. I render them in latex and copy-paste.

Jasper Habicht

@cfr Ha! That's the way to go!

cfr

@Skillmon not that I know of. sorry.

Skillmon

@cfr that's the way I produce the visible space glyph thingy. I open LaTeX and enter \verb*| | and copy over the symbol :)

@cfr more seriously: Have you tried the compose button?

@cfr at least on my machine ♫ss yields ß (even though I could type it directly)

cfr

@Skillmon well caps-lock is compose, but I don't know what combination I'd need for a pound sign. € is fine. which probably means I'll be executed for treason or something these days.

ß

Skillmon

@cfr I meant the "ß" problem

Pounds should be ♫-l for me (that's compose+hyphen+L)

cfr

7:01 PM

@Skillmon I realise :-). and thanks. though I do need pound signs slightly more often.

Skillmon

@cfr see above :)

cfr

£

David Carlisle

@cfr they are never equal

cfr

@Skillmon diolch yn fawr!! is there a list somewhere? I just guess.

@DavidCarlisle \ifthenelse says otherwise.

David Carlisle

@cfr I get

> left:1{} one.
> right:\foreignlanguage {british}{\bbl@restore@actives 1 a}.
> left:3{} three.
> right:.
> left:5{} Implementation.
> right:.

Skillmon

7:03 PM

@cfr a quick google search yields unix.stackexchange.com/questions/248017/…

David Carlisle

from

\cs_new_nopar:Nn \__cfrlm_doc_lreq: {
  \group_begin:
\let\protect\noexpand
  \tl_show:e{left: \firstleftmark}
  \tl_show:e{right: \firstrightmark}
\tl_if_eq:eeTF { \firstleftmark } { \firstrightmark } {} {\firstrightmark }

@cfr ^^

which ones would you expect to be equal?

cfr

@DavidCarlisle they are equal in the change history and index. \ifthenelse gets that right. but I can't figure out an expl3 alternative which gives me the right answer in the index and the main document.

David Carlisle

@cfr which are equal, the values shown all look different, so I don't know which you expect to test equal

cfr

@DavidCarlisle on the last page you should get 'Index : : Index' i.e. the middle is blank because the marks are both 'Index' (at least once expanded).

David Carlisle

@cfr oh do I need to run makeindex etc to actually build an index (I just ran pdftlatex a couple of times and tested the marks, as above)

@cfr so my last page is 5 implementation

cfr

7:09 PM

@DavidCarlisle I ran makeindex and the glossary thing. that's why my so-called minimal example contains so many \DescribeMacros I needed the index to run to a second page.

@DavidCarlisle should be 6.

David Carlisle

@cfr let me try again aafte makeindes (using is gind.ist I assume)

> left:Index.
> right:Index.

cfr

@DavidCarlisle makeindex -s gind.ist ... ; makeindex -s gglo.ist -o <file>.gls <file>.glo.

@DavidCarlisle exactly.

David Carlisle

@cfr yes that's what I get

cfr

so I can get the correct result before the index, but am stuck with three rather than two 'Index'. or I can get just two 'Index', but the result before that is wrong. but I can't get the right result in both places without \ifthenelse.

David Carlisle

cfr

7:16 PM

@DavidCarlisle yes. that's what I get with the mwe. but the header on page 3 is wrong. I can change the code so the header on 3 is correct, but then I get an extra 'Index'. or I can use \ifthenelse and I get both the right header on 3 and only two 'Index'.

that is, the header on 3 should be 1 one : 1 a : 2 two and not 1 one : british1 a : 2 two. does that make sense?

Ulrike Fischer

@cfr untested but you can probably do \protected@edef\blubA{\firstleftmark}\protected@edef\blubB{\firstrightmark} and then compare \blubA and \blubB (assuming that ifthenelse does this expansion)..

@cfr but perhaps it would be better not to use extramarks but the new marks code.

Skillmon

@cfr hm, my quick google tells me it depends on the used desktop environment which compose combinations are available. Though I think your keyboard layout might have additional impact. So the definitive list is hard to find.

cfr

@UlrikeFischer I would love to use the new marks code but it doesn't work with ltxdoc. it works beautifully until you get to the multicol bit, but then it is ... interesting.

David Carlisle

@cfr I get the same vaues with ifthen an tl_e

@cfr I used

\cs_new_nopar:Nn \__cfrlm_doc_lreq: {
  \group_begin:
\let\protect\noexpand
%  \tl_show:e{left: \firstleftmark}
%  \tl_show:e{right: \firstrightmark}
[\tl_if_eq:eeTF { \firstleftmark } { \firstrightmark } {tl-no} {tl-yes-\firstrightmark }]
\group_end:
[\ifthenelse { \equal { \firstleftmark } { \firstrightmark } } {ifthen-no} {ifthen-yes-\firstrightmark }]
}

Then all the pages that use that header show the same branch for tl_if_eq and ifthenelse

cfr

@DavidCarlisle I definitely don't get the same.

@DavidCarlisle let me try that

(but it should be the same?)

David Carlisle

7:26 PM

@cfr only the last page Index has marks that test equal and they test equal with expl3 and iftenelse

cfr

@DavidCarlisle yes, but the mark comes out incorrectly on page 3. I get e.g. 1 one : [tl-yes-british1 a][ifthen-yes-1 a] : 2 two at the top of 3. i.e. expl3 gives me british1 a. ifthenelse gives me 1 a.

@DavidCarlisle i included the index because I can solve the header problem on 3 in isolation, but in doing so I get the wrong result in the index.

it's the same branch, but something is different about how it comes out.

David Carlisle

@cfr but you only use it in the test (and you get not-equal with either test) I typeset it for debugging but using either test the test is not-equal so you use \firstrightmark ohhh I see

You probably want [\tl_if_eq:eeTF { \firstleftmark } { \firstrightmark } {tl-no} {\let\protect\relax tl-yes-\firstrightmark }] so that you restore \protect for typesetting and just have it \noexpand for the test

so without the debugging \tl_if_eq:eeTF { \firstleftmark } { \firstrightmark } {} {\let\protect\relax \firstrightmark }

@cfr or move the \group_end: inside both branches of the test (which is what ifthenelse does)

@cfr vvv

\group_begin:
\let\protect\noexpand
\tl_if_eq:eeTF { \firstleftmark } { \firstrightmark }
{\group_end:} {\group_end: \firstrightmark }

cfr

@DavidCarlisle diolch!!

[might be ddiolch!! in some contexts.

David Carlisle

7:42 PM

@cfr beio pwy bynnag a ddywedodd wrthych am ailddiffinio amddiffyn

cfr

@DavidCarlisle I can guess what you intended to say if I put it back through google translate.

cfr

8:00 PM

@UlrikeFischer this also works now I remembered I removed \makeatletter when minimising the example. diolch!

David Carlisle

@cfr I am sure my Welsh was perfect, if you needed to use google translate to help you read it, I can't help with that, sorry. @PauloCereda has a similar problem with Portuguese as I recall.

Rmano

@cfr tuttle.github.io/python-useful/compose-key-cheat-sheet.html this is quite good, it groups similar things

A pity there's no compose on the tablet... :-(

Skillmon

8:32 PM

@Rmano but it looks quite incomplete.

Skillmon

8:43 PM

@JosephWright was there a recent-ish change in l3doc regarding \marg? For me it changes fonts in one of my documentations. I have some code to turn \meta into always using \ttfamily, but somehow \marg uses \rmfamily now.

Hm, looks like it used to use a nested \meta but was changed to use \__codedoc_meta:n instead.

cfr

@Rmano diolch. it does look a bit incomplete. and I think @Skillmon said it varies. one very annoying thing about my keyboard is that the key I need for € can't also be used for compose. not that I particularly miss caps lock most of the time.

Skillmon

@cfr hm, on my layout Compose is activated with Mod3+Tab (and Mod3 is what you'd call the CapsLock key)

cfr

@Skillmon I changed caps lock to work as compose. configuring keyboards in linux seems to be something of a dark art. since none of the layout keyboard types have ever corresponded to any actual keyboard I've owned, it always seems somewhat random. and kde always claims not to recognise settings x11 accepts, but then overwrites the rest anyway. and also my mouse, touchpad and pointer-thing settings. kde is just determined to fight about everything.

Skillmon

8:59 PM

@cfr yes, I tried to setup Neo (an optimised German setup) on my wife's laptop using KDE. It worked reasonably well, but the keyboard shortcuts were somewhat messed up (they are hardcoded to physical buttons in KDE, so whichever default keyboard layout I specify wins, and now locking is on Super+E because that's where the L is in Neo).

cfr

@Skillmon capslock + tap makes a sound, takes me up the transcript and then starts tabbing through, I guess, all the clickable items.

Skillmon

@cfr And I don't mangle with the layout myself, I just let the Neo community do their splendid job (though there are some applications which don't like the entire range of the layout, most notably the menus of my WM don't accept layer 4 and 6)

@cfr well, that's Neo doing its magic for me :)

cfr

@Skillmon you have to install the kcm module and change the kde settings to match the config settings for x. however, if kde's kcm doesn't offer the settings you want, things get interesting (and sometimes kde doesn't seem to recognise its own settings either). so for pointing devices I have a shell script which runs to override kde at key points or on demand.

Skillmon

@cfr also for the composition maps: You can take a look whether ~/.XCompose exists. If it does you can look up available compose sequences there.

cfr

@Skillmon no such file, sadly. :(

enkorvaks

11:02 PM

@Rmano "A as is Aisle; G as in Gnome; H as in Hour; K as in Knife; P as in Pterodactyl" I see no problems with a dictionary being in phonetic order :-)

Transcript for

$TeX - LaTeX$ TeX, LaTeX and Friends