12:04 AM
quack

2 hours later…
1:34 AM
@CarLaTeX hello! I have tried with the 2 answers not knowing how to deal with the given commands
For example if I take the accepted answer with this code I get:
\documentclass{article}

\usepackage{tikz}

\usetikzlibrary{fit,positioning,hobby}

\begin{document}

\begin{tikzpicture}
\node (1) at (0,0) {$1$};
\node (2) at (-1,1) {$2$};
\node (3) at (1,1) {$3$};
\node (4) at (-2,2) {$4$};
\node (5) at (-1,2) {$5$};
\node (6) at (1,2) {$6$};
\node (7) at (2,2) {$7$};
\node (8) at (-4,3) {$8$};
\node (9) at (-3,3) {$9$};
\node (10) at (-2,3) {$10$};
\node (11) at (-1,3) {$11$};
\node (12) at (1,3) {$12$};
\node (13) at (2,3) {$13$};
\node (14) at (3,3) {$14$};
Here I do not know how to add more fit nor how to combine them to include all numbers involved
If I use the 2nd answer I get:
\documentclass{article}

\usepackage{tikz}
\usetikzlibrary{fit,positioning,calc}

\begin{document}

\begin{tikzpicture}
\node (1) at (0,0) {$1$};
\node (2) at (-1,1) {$2$};
\node (3) at (1,1) {$3$};
\node (4) at (-2,2) {$4$};
\node (5) at (-1,2) {$5$};
\node (6) at (1,2) {$6$};
\node (7) at (2,2) {$7$};
\node (8) at (-4,3) {$8$};
\node (9) at (-3,3) {$9$};
\node (10) at (-2,3) {$10$};
\node (11) at (-1,3) {$11$};
\node (12) at (1,3) {$12$};
\node (13) at (2,3) {$13$};
\node (14) at (3,3) {$14$};
Here I do not know how to trim the protruding parts nor how to make the rounded edges
Do you want to move to the Q&A site?

4 hours later…
5:20 AM
\documentclass{article}

\usepackage{tikz}

\usetikzlibrary{fit,positioning,hobby}

\begin{document}

\begin{tikzpicture}
\node (1) at (0,0) {$1$};
\node (2) at (-1,1) {$2$};
\node (3) at (1,1) {$3$};
\node (4) at (-2,2) {$4$};
\node (5) at (-1,2) {$5$};
\node (6) at (1,2) {$6$};
\node (7) at (2,2) {$7$};
\node (8) at (-4,3) {$8$};
\node (9) at (-3,3) {$9$};
\node (10) at (-2,3) {$10$};
\node (11) at (-1,3) {$11$};
\node (12) at (1,3) {$12$};
\node (13) at (2,3) {$13$};
\node (14) at (3,3) {$14$};
@manooooh Try this ^^^ but it would be better using a tikzmatrix for node positioning or some package for trees

3 hours later…
8:04 AM
I wrote an answer and between me writing it all up and posting it, it has been deleted :(

8:45 AM
@Skillmon The OP was mean

@Skillmon oh no
quack

9:36 AM
@DavidCarlisle 'ello
@JosephWright John Cleese is 80 today!

9:59 AM
@DavidCarlisle I see you are having with with iftex
@DavidCarlisle I've been trying to work out the WEB for \input changes, but ... I need a break from that ... it's frustrating as I have no real idea what I'mdoing

10:41 AM
@JosephWright can we do "loadable in initex" tests in l3build?
@JosephWright I was planning to get back to filename handling this morning (not sure I can get it to work svg include with just kernel changes, I gave up last night too tired, but I suspect the catcode 12 svg extension will need changes in the package, but we'll see...

@DavidCarlisle ooh

@DavidCarlisle We'd need to find a way of running the tests ... probably have to forgo regression-test.tex
@DavidCarlisle I guess: what did you have in mind?
@DavidCarlisle Got it

@JosephWright my current plan for oberdiek (I think, but I keep changing it) is to block out the generation of if[pdf|vtex|luatex].sty but leave the sources there and include them in hobsub-[hyoerref|generic] but just not generating hobsub-* at all is tempting, nothing in texlive uses it except hyperref
@JosephWright something that meant I didn't get "loading in initex is broken" reports just after I commit something that says "all tests pass" (that's as far as my planning for that feature got:-)

11:05 AM
@DavidCarlisle Working with IniTeX is pretty unusal, relly

@JosephWright yes (I'm not sure I needed it here except that the only thing I needed from plain tex was catcodes of {} and one instance of \newif so it seemed simpler not to use newif and say it worked in initex, actually it turns out it wasn't quite that simple but anyway, done now...

11:21 AM
4
@PauloCereda I had this in my talk yesterday ^^^

@UlrikeFischer oooooh

@UlrikeFischer Slides going on the team site?

@UlrikeFischer yay
@JosephWright bah

@JosephWright hm, good question. Someone would have to check the pictures. Beside this my slides have a tendency not to contain much ;-).
@JosephWright but I'm boarding now in Frankfurt, so later ...

@UlrikeFischer have a safe flight! <3

11:36 AM
@moewe In tex.stackexchange.com/a/330767, giveninits is mentioned, but it is not (or no longer) documented in biblatex's documentation as far as I can see. Might I ask you to you comment on this and, if necessary, adapt your answer? Thank you in advance.

11:58 AM
@PauloCereda New version of a flight accident: the bus had to make a sharp brake as it nearly crashed into a plane ...

2 hours later…
1:51 PM
Concerning tex.stackexchange.com/questions/503000/… : hidelinks=true is widespread (g**gle for it!). Has there been a change in the syntax recently?

2:23 PM
@UlrikeFischer oh my

@MdAyq not as far as I know (and the master copy is on this machine:-) and there has been no change since at least texlive 2016:
\$ grep hidelinks /usr/local/texlive/*/texmf-dist/tex/latex/hyperref/hyperref.sty
/usr/local/texlive/2019/texmf-dist/tex/latex/hyperref/hyperref.sty:\DeclareVoidOption{hidelinks}{%
@JosephWright Heiko's tests aren't fair:-) I ran l3build check in oberdiek after having blocked out if*.sty just to check nothing broke only to find he'd set the catcode of " to invalid character :(

@DavidCarlisle That's the whole business about him wanted to assume catcodes of only letters, numbers, space, backslash (braces?)

but actually " being active isn't that unlikely so I made tex.enableprimitives ("IFTEX@" safe anyway. I'd originally thought I'd be modern and use \detokenize to avoid having to mess with local catcode setup, now I have a local catcode setup to enable detokenize...

@CarLaTeX well, his question was a do-it-for-me, and I thought the answer was easy enough to write it down quickly. But it got deleted less than a minute before I wanted to post :(

2:40 PM
@MdAyq I tried google but didn't find any uses of hidelinks=true, do you have an example?

@DavidCarlisle Some people are concerned that the links are highlighted in print, too, so they use hidelinks=true.

@DavidCarlisle You find it in eso.org/~jagonzal/ADASS-2016/ADASS2017_author_template/… , for instance. I've also been told that it's in some SIAM class (but I have not verified it myself). http://www.google.com/search?client=ubuntu&channel=fs&q="hidelinks%3Dtrue" returns over 783 results for me.
@DavidCarlisle There is a typo above. To be precise, http://www.google.com/search?q="hidelinks%3Dtrue" returns exactly 783 results for me.

2:59 PM
@MdAyq hmm that's from 2014 I have no change log going back that far (we picked up support in 2016.
@Skillmon point is that hidelinks=true is an error the syntax is [hidelinks] (it is declared as a void option not a boolean, defaulting to true)

@MdAyq you could open a change request at github.com/latex3/hyperref/issues with a test file, could look at it later, thanks.

Meanwhile, I avoid using hyperref at all costs. I am a happy duck. :)

@MdAyq well there has been imho no syntax change, but the message is only a warning and hyperref is doing the right thing. So probably people didn't look in their log ....

@UlrikeFischer ah....

3:04 PM

@PauloCereda a policy followed by 99.99995% of all TeX users

@DavidCarlisle ooh rounding

@PauloCereda I didn't want it to look like I had just made up a random length string of 9s

@DavidCarlisle I'm not sure what's the right thing to do: it may be either to change the syntax of the option in hyperref or to change some instances of its usage. So, I won't issue a ticket myself. But, if anyone believes he/she/it is doing the right thing, they should feel free...
@UlrikeFischer Thanks, good to know!

@DavidCarlisle ooh it's like saying no in German in an angry tone

3:06 PM
@DavidCarlisle yes, but now the editors interfere: they create lists of these warnings and users suddenly worry ...

@UlrikeFischer cunning way to make the if.. tests pass in oberdiek without having to assume the package works with every character declared invalid
diff --git a/testfiles/accsupp-test1.lvt b/testfiles/accsupp-test1.lvt
index 4fda36d..b14dfa7 100644
--- a/testfiles/accsupp-test1.lvt
+++ b/testfiles/accsupp-test1.lvt
@@ -25,7 +25,8 @@
\repeat
}
\def\Test{%
-  \RangeCatcodeInvalid{0}{47}%
+  \RangeCatcodeInvalid{0}{33}%
+  \RangeCatcodeInvalid{35}{47}%

@PauloCereda I often read the log, and sometimes I do discover issues (e.g., lost chars or superfluous chars) that would have missed my attention otherwise. The down side is that looking into a log eats up my time.

@MdAyq I also ready my logs. :)

@DavidCarlisle oh, you're right, my bad.

@UlrikeFischer But since this behavior (doing the right thing) seems undocumented, I wouldn't vouch to rely on it.

3:12 PM
@MdAyq ? that's was message is saying: it ignores the value.
@DavidCarlisle why do the test now needs an valid "?

@UlrikeFischer because ifluatex is using the one from iftex and that guards against makes " safe using \detokenize so active " is safe but an invalid character is not but I decided I don't care, any document that has made characters invalid should take care of protecting things if loading other packages
I don't want iftex.sty to be like another package where more than half of the code is saving, setting then restoring the catcode of every non-letter character.
@MdAyq @UlrikeFischer we could make hidelinks a boolean, with default true then hidelinks will work as before and hideliknks=true won't warn, but (say) hidelinks=false which is currently same as hidelinks would change value and hidelinks=yes which is currently the same as hidelinks would become an error so you have to balance the benefit of not getting a warning with the cost of possibly changing existing documents.
grrrrrr
(./hobsub-generic.sty

! Package ifvtex Error: Name clash, \ifvtex is already defined.
@UlrikeFischer @JosephWright not building hobsub gets more and more attractive

3:29 PM
@UlrikeFischer Oh, are saying that the effect of "hidelinks=true,..." is equvalent to "hidelinks=false,..." is equivalent to "hidelinks,..." ?

@MdAyq yes

@DavidCarlisle Thx!

@DavidCarlisle I will check if hidelinks=false makes sense. Do you know if luatex should actually be able to use \pdfextension glyphtounicode?

@UlrikeFischer no idea!

https://www.google.com/search?q="hidelinks%3Dfalse"+hyperref returns 34 results for me. So, some people got an output which they had not intended to get. If you plan to turn hidelinks on and off in one document in calls to \hypersetup, then, turning hidelinks into a boolean with an explicit value would be a good idea. If you don't plan this, keeping the option valueless as it is would be a good idea.

3:41 PM
@DavidCarlisle I found an answer: tex.stackexchange.com/a/404575/2388

@PauloCereda -- Wow! I've done that. Thanks for the memory.

@UlrikeFischer must be true

@DavidCarlisle in the current implementation hidelinks=false doesn't make sense.

3:59 PM
@UlrikeFischer let's just forget it then. Meanwhile I broke hyperref anyway.
! Extra \fi.
l.2947 \fi
@UlrikeFischer ^^

@DavidCarlisle some change in \ifHy@DviMode or earlier?

@UlrikeFischer somewhere around there yes, so either the ifvtex.sty emulation in iftex isn't right or my dis-entangling hobsub-hyperref broke something or ....

problably the test just before this?
@DavidCarlisle how can I test it? Install oberdiek and then try to load hyperref?

@UlrikeFischer you could come here and look over my shoulder.
@UlrikeFischer I have a pile of uncommited changes in oberdiek to not build if*tex.sty and just for once trying to get the tests to pass before committing them.
@UlrikeFischer oh could you test this:
\documentclass{article}

\usepackage{hyperref}
\begin{document}

\end{document}
but first change lines 105-8 of hyperref.sty to
%\IfFileExists{hobsub-hyperref.sty}{%
% \RequirePackage{hobsub-hyperref}[2011/01/30]%
%}{}
@UlrikeFischer ah it's an iftex bug it works if I uninstall the new iftex, should find it easily now
\ifvtexdvi=undefined.
(@UlrikeFischer)

4:23 PM
@DavidCarlisle where is that?
On other news: I just updated my context to check the problem with the non-ascii file name (which fails there too), and it did install luahbtex.

@UlrikeFischer in the new ifvtex.sty I only defined the vtex-specific \if... if it was running in vtex, which means that hyperrefs nested ifs go wrong , if you load vtex.sty but not in vtex it needs to define them all to \iffalse, yet another iftex checkin coming up...

ah, the \ifvtexdvi was the problem.

@UlrikeFischer yep it wasn't defined but it was skipped from an outer \iffalse so no error there but then all the \fi matching after that is wrong

1 hour later…
5:45 PM
@MarcelKrüger this is breaks in harf mode for me with a table index is nil:
\documentclass{article}
\begin{document}
\font\test={file:cmunrm.otf:mode=harf;}
\test

abc123
\end{document}

arrrrrrrrrrrrrrrrrrggggggggggggggggggggggggghhhhhhhhhhhhhhhhh

@DavidCarlisle ?????????????????????

@UlrikeFischer wasn't my message clear enough?

@DavidCarlisle only wondering which package broke ;-)

I set off ./mkctan in oberdiek about an hour and a half ago and it was just getting to the final stage and
! LaTeX Error: File ifluatex.toc' not found.

5:58 PM
@DavidCarlisle ;-(. Did you drop hobsup?

@UlrikeFischer yes I was going to keep it but if you load ifxetex then hobsub (as hyoerref did) then because ifxetex loads iftex and defines \ifvtex then hobsub's internal version of ifvtex package throws a fatal error as \ifvtex is already defined. At that point I decided it had to go. I'll make a basic stub that just does a list of \RequiePackage I think, but I was building oberdiek with a modified version of hyperref that had the iffilexists{hobsub} commented out

@DavidCarlisle good. It should gives us a chance to split a bit.

1 hour later…
7:28 PM
@MarcelKrüger perfect, works now ;-)

@UlrikeFischer Fixed. Actually a interesting problem: If a font assigns an explicit Unicode mapping for the glyph at GID 0, HarfBuzz reports that the corresponding Unicode value is mapped, but later acts as if the unicode value were not defined and using GID 0 as a fallback. I think this is either a HarfBuzz or a luaharfbuzz bug, but the HarfBuzz documentation is ot clear enough to decide who to blame.
@UlrikeFischer We even get good tounicode mappings there.

@MarcelKrüger I was just trying what tounicode mappings harfbuzz created, when I got the error ;-)

If only all harf problems were that easy to fix...

@MarcelKrüger imho it works very well already. Luigi said that they are considering your two suggestions - for the callback there is the question of compability. He also said that they will be a release shortly, and probably another one in februar, so we can run tests in between.

8:12 PM
If the Harfbuzz + LuaTeX merge thing in process?

8:23 PM
@FaheemMitha yes.

Ok. Thanks.

@UlrikeFischer It mostly works, but given the Travis problems I don't really trust it. Also I get a similar thing where it just freezes when I'm testing Graphite fonts. I'm still trying to figure out where that problem is coming from.

@MarcelKrüger it is not easy to debug travis. Which binary it is using there?

The LuaHBTeX 1.11.1 build from texlive-contrib. I tried the same binary but still couldn't reproduce.

@UlrikeFischer checked in oberdiek...

8:29 PM
@DavidCarlisle I'm running the tests.

@UlrikeFischer they should pass except the askinclude ones (which fail over .aux.aux filename breakage)

@MarcelKrüger I will make a try with a simple test file which doesn't use harf mode.

@UlrikeFischer It already runs the non-harf tests successfully, so I'm pretty sure that it is a harf thing.

8:45 PM
@MarcelKrüger I removed the ini from the checkopts and now it no longer freezes but simply errors as the pdf is not there. But locally it works.

8:57 PM
@CarLaTeX THANK YOU SO MUCH IT WORKED <3. It took me like an hour to produce this:
I used:
\draw[thick, red, rounded corners] ([xshift=0pt, yshift=2pt]12.north west) --
([xshift=-2pt,yshift=-2pt]6.south west) --
([xshift=0pt, yshift=-30.5pt]14.south east) --
([xshift=0pt, yshift=-2pt]14.south east) --
([xshift=2pt, yshift=-2pt]15.south east) --
([xshift=2pt, yshift=2pt]15.north east) --
([xshift=0pt,yshift=2pt]14.north west) --
([xshift=14pt, yshift=2pt]7.north east) --
([xshift=-14pt, yshift=2pt]7.north west) --
([xshift=0pt, yshift=2pt]12.north east) --
cycle;

\node[text=red] (LBL) at ([xshift=-20pt, yshift=-40.5pt]14.south east) {$$A\subseteq\mathbb{N}$$};
But I enjoyed the challenge. @CarLaTeX if you know a better proposal in changing shift values and/or drawing the lines in other positions, I will be grateful. Thank you!

9:42 PM
@UlrikeFischer -cmdx, as the first option I half remembered there was some extra option somewhere:-)
@UlrikeFischer Ulrich must mean you:-)

@DavidCarlisle ? I'm lost, did I miss something?

@UlrikeFischer Akira replied re German filenames in luatex on tl list

@DavidCarlisle hm, it works but it really needs to be the first option. But I wonder what this is for an incompability with context.

@UlrikeFischer seems very odd if context works in linux with utf-8 files and presumably can work over miktex's luatex

9:58 PM
@DavidCarlisle yes, and I don't think that this will do. And unicode engine like luatex should work without such extra options.

@UlrikeFischer i wonder if the luatex issue he's trying to avoid still applies with current luatex/context?

@DavidCarlisle not sure. I will write Luigi, he suggested that I test context and as it fails there too ...

@UlrikeFischer this person seems to have tested it last year tug.org/pipermail/luatex/2018-September/006954.html

@DavidCarlisle ah yes, I remember. I found also the texlive thread.

1 hour later…
11:32 PM
@MarcelKrüger it is getting better. It seemed to have hanged because of the missing setspace.sty. And I think now NotoSerifCJK-Regular.ttc is missing.