quack

@PauloCereda Hi

@JosephWright ooh another duck

morning

@mag 'ello (")>/

I just had to find out that you can't condition \hhline 's with \ifthenelse constructs

and that kind of threw my whole document design off

@mag ifthenelse is not expandable, but there are alternatives. Ask a question with an example on the main site.

@UlrikeFischer the parcel has been... dispatched. :)

@UlrikeFischer Probably have to do that. In essence what I'm making is a table in the footer of a document that's only supposed to be visible on the last page of the document

and that table contains a \hhline because I need it

Hi mr. velociraptor

ooh Akira-san is here!

@AkiraKakuto hi!

@mag If the whole table just needs to be in the foot, why do you need a conditional hline?) but you need to avoid \ifthenelse anywhere that you need expansion, actually I'd avoid it altogether, despite the fact that the package doc claims I wrote it.

@PauloCereda more dinner

@DavidCarlisle oh no

@PauloCereda hi. I'm infrequently here to study recent news.

@AkiraKakuto cool! :) Thank you very much for the amazing work you've done with TeX!

@DavidCarlisle it's a footer table that represents the whole foot of the document. Due to strict positioning requirements I fit the footer around a table. It has an image, a cert logo and a signature field in it, as well as IDs and other text. the signature field is supposed to be visible only on the last page. I use cellcolor and a tabularx in with it to color the background of the signature field gray, as per specs. Because of this, a \cline is insufficient for drawing clear lines across the tabl

as I need different line colors within a single line.

you would probably be less-then impressed of all the workarounds and dirty hacks I had to do to get TeX to play well with a automatic-report generation system

@mag the question is if you want to suppress the whole content of the table or only parts of it on the last page.

@mag I don't really follow but \ifnum\value{page}=10 \else \hline...\fi will work to put an hline just on page 10

@UlrikeFischer part of the table

what I tried is \ifthenelse{\equal{\thepage}{\pageref{LastPage}}}{\hhline{~>{\arrayrulecolor{\tblbgcl}}->{\arrayrulecolor{black}}-~}}{}

The document is dynamic and can have 1 to n pages when printed

@DavidCarlisle everyone is so eager to enable tests - isn't that nice ;-)

@mag well this won't work, lots of unexpandable stuff here. You should really better ask on the main site with an example.

I noticed it doesnt

Making a working example is nigh on impossible for my use case though, so I'm barred from asking on the main page

it's a dynamic report system using proprietary data (even the formatting is) and the document is long and intertwined.

what the final file looks when it is compiled is entirely different from the template I'm editing, and there are 2 subsystems doing code replacement in the code before it gets compiled, and an additional converter it is run through, triggered by specific comments in the templates. It's all rather a mess tbh

@mag no it isn't. You want to suppress a line in a tabular in the footer in dependance of the LastPage reference - it is easy to make a small example with this scenario.

the line that exists is dynamic also, however

@UlrikeFischer I forgot to put a Blame Ulrike, who names all her files using ß

@DavidCarlisle that's boring, does she also use ẞ?

@mag you can easily make an example, just start from the line you quote above an dwrapit in the minimum required to make it a complete document with more than one page to show the effect.

@Skillmon I think she wanted a β but it's so hard to find anything on a German keyboard

@DavidCarlisle ;-). I just wanted to check if luatex works now with "grüße".

@UlrikeFischer ooh

@DavidCarlisle β is on my keyboard...

@Skillmon is z in the right place too?

@DavidCarlisle of course: neo-layout.org

@Skillmon eek everything moved:-)

@DavidCarlisle no, "m", "," and "." are fine.

@Skillmon mmmm.

@DavidCarlisle we windows users are still excluded from \includegraphics{grüße} with lualatex in texlive. It would probably pushing luck to much to ask for an engine update ;-).

@UlrikeFischer well, you windows users are doomed in the first place...

@UlrikeFischer you could reply on list and ask, although as that doesn't require a new binary name and new format config I suppose it's easier as you can (I assume) just grab a luatex.exe from Akira's site and drop it in?

@UlrikeFischer but if we get a new luahbtex in then presumably that will have the new filename code and you can use that.

@DavidCarlisle I have the new binary too and with it it works fine (and perhaps you can also get it with tlcontrib, I didn't try), but it is as with luahbtex: it is more difficult for the normal user.

@UlrikeFischer yes but what I mean is, people who want to test things can use lualatex-dev and (hopefully) that will have the new filename handling as well as harfbuzz

@DavidCarlisle yes, let's look how long it will take and who is first luahbtex or latex-dev ;-)

@JosephWright if I do l3build ctan in latex2e/iftex then it leaves an iftex.pdf in the top level directory is that expected? it leaves 3 copies in build/* as well but I have build in .gitignore but git status is moaning about iftex.pdf (I could ignore it as well but...)

@DavidCarlisle Yes the script copies PDF files from build to their 'home' as they are likely useful and not 'build artefacts'. Most of the repos have *.pdf in the .gitignore

@JosephWright ah, I was wondering about such a pdf two days ago too. What is the "home"?

@UlrikeFischer docfiledir

@JosephWright I was worried that stray .pdf that are at the top level (eg texput.pdf from interactive experiments) would get zipped up with no warning but apparently not as you zip up in a clean area than copy, so OK this seems fine, I'll gitignore the pdf, thanks

@DavidCarlisle Yes, all zipping is done inside build/distrib

@DavidCarlisle Same for any other stray files, hence explicit copy to a staging area

@UlrikeFischer @JosephWright you might want to proof read the last section of iftex.pdf that I just added, hope it's diplomatic enough...

@DavidCarlisle Looks fine to me

@DavidCarlisle it is a bit unclear where you have to use the option internalonly

@UlrikeFischer er anywhere that works:-) You can also use extravagant versions like \usepackage[internalonly=/ifpdftex]{scrbase} (which I use in one of the test files, but I didn't want to say too much, basically it should work on the \usepackage and it will work on the \documentclass for komascript classes, but if scrbase is being included indirectly via some other class it depends a bit on how it is passing options on (as it is using key value options it all depends..)

@UlrikeFischer is it known issue that the harf mode in luahblatex is broken with Polyglossia? I've looked in issue trackers of both Polyglossia and Luaotfload and couldn't find any report.

@michal.h21 do you have an example? And what means "broken"? Wrong script or an error?

It produces error

\documentclass[12pt]{article}

\usepackage{fontspec}
\setmainfont{Latin Modern Roman}[RawFeature={mode=harf}]

\usepackage{polyglossia}

\begin{document}

Hello world

\end{document}

@DavidCarlisle yes, but it sounds as if one should/could use it as an iftex option. Perhaps "use the scrbase option ..." or the "use the KOMA option".

...exmf-dist/tex/luatex/luaotfload/luaotfload-auxiliary.lua:444: attempt to inde
x a nil value (field 'hbshared')
stack traceback:
	...exmf-dist/tex/luatex/luaotfload/luaotfload-auxiliary.lua:444: in upvalue 'ge
t_hbface'
	...exmf-dist/tex/luatex/luaotfload/luaotfload-auxiliary.lua:529: in field 'prov
ides_language'
	.../texlive/2019/texmf-dist/tex/latex/fontspec/fontspec.lua:55: in field 'check
_ot_lang'
	[\directlua]:1: in main chunk.
\lua_now:e #1->\__lua_now:n {#1}

l.8 \begin{document}

@michal.h21 I will look at it.

@UlrikeFischer OK I'll adjust..

 \verb|\ifpdftex| and \verb|\ifVTeX| with a different syntax.  If you
-use the option \verb|internalonly| then it will not define these and
+use the \textsf{scrbase} option \verb|internalonly| then \textsf{scrbase} will not define these and
 the definitions as described here will take effect. This is

@UlrikeFischer ^^ ?

@DavidCarlisle yes ok.

@UlrikeFischer thanks

@UlrikeFischer thanks

@michal.h21 I think there is a typo in luaotfload-auxiliary. get_hbface should return tfmdata.hb.face, but @MarcelKrüger should better confirm this.

@DavidCarlisle ooh

@michal.h21 no actually it should be return tfmdata.hb.shared.face. <-- @MarcelKrüger

@UlrikeFischer this fixes it. there is also another bug - missing defintion of stringupper, which should be local stringupper = string.upper I guess.

@michal.h21 Thanks. @UlrikeFischer I pushed both fixes to dev. I probably can push a base version of the script stuff later today, maybe we can push a bugfix release then with both changes?

@MarcelKrüger yes, I can do it in the evening or tomorrow.

Also if this code was broken then there still is a fontspec problem. If fontspec would handle Renderer = HarfBuzz correctly, this should have been detected by our tests.

@MarcelKrüger imho the error only happens if \fontspec_if_language:nTF is used.

@UlrikeFischer It should also trigger if \fontspec_if_script:nTF or \fontspec_if_feature:nTF is used. This should basically happen whenever fontspec loads a font, so if that doesn't happen trigger the bug it means that the initial load does not pass mode=harf. That's a problem for huge fonts because then the fontloader still has to load the font, even if the font is never used by the fontloader.

@MarcelKrüger looks as if the typo was a useful error ;-)

@MarcelKrüger can we add (or is there) some debug mode which prints messages like "font loaded in harf mode" and "font loaded by fontloader" which would help Will to debug this?

Can we assume any bugs reported to the fontspec bug tracker are @WillRobertson's fault?

@DavidCarlisle there is catcode problem in shellesc when used with lualatex:

\documentclass{article}
\showthe\catcode`\" %12
\usepackage{shellesc}
\showthe\catcode`\" %8

\begin{document}
\end{document}

@UlrikeFischer that's what I was referring to in my last comment, see github.com/wspr/fontspec/issues/389 I blame @WillRobertson

I suspect a tools update is coming up.... CTAN will be so happy that @JosephWright and I are keeping them from getting bored.

@DavidCarlisle oh, I came from the texlive list: it also broke \verb|abc"abc| if minted is loaded.

@UlrikeFischer sure it breaks any use of " (ah there is a bug on minted that I had main for (I watch it) but haven't looked I had better check...

@DavidCarlisle the fix works fine.

Hello!

Is somebody a Q in sage(math)?

:(

@cis What are you asking?

@TorbjørnT. I want to implement the Collatz-series

@cis I don't even know what that is. But you're setting i=x at the start of the loop, so it's always 19, and you get the same output five times.

@TorbjørnT. Yes, I see.

@cis if your program doesn't stop, you'll be famous.

@DavidCarlisle I am sure, I will get the Fields Medal... :()

@BrainStone you could update your texlive (oberdiiek and iftex (at least) are not the latest

Hm. Ok. That's weird

I'll look into that

@BrainStone well I only noticed as I uploaded them to ctan within the last couple of days

I created that image just today

Maybe my installation method is not up to date

@BrainStone or maybe your ctan mirror is a day behind, it takes two or three days for all mirrors to catch up

@BrainStone but more generally it is very rare to have a difference because of miktex.texlive it is almost always due to different installed package versions, so compare the log files and see which package versions differ

@BrainStone if you would show the preamble I could try, but I will not assemble it from the log ...

@UlrikeFischer gist.github.com/BrainStone/1ea904a31c12ea42ca61759fac9fffa3

@DavidCarlisle weird. I just updated my miktex installation yesterday

That's how I install texlive: gist.github.com/BrainStone/bbe92798c432883a264afba321b3176f

@BrainStone that is not the preamble from your log. I don't have your local fonts, without them it compiles fine. Remove the font setup, compile only with lualatex, not the latexmk, and if it fails show the real log, not the terminal output.

Ahhhh

That's what you mean

My bad

@UlrikeFischer How much of the log do you need?

I mean I can upload it in full

@BrainStone the full log. But it should be from the tex file so that it makes sense to compare with mine.

@UlrikeFischer gist.github.com/BrainStone/8b3a7564e0f000e9690f5c16a4a575df here you go

That's the xxx.log file of my xxx.tex file I tried to compile with lualatex directly

@BrainStone the tex you posted didn't load inputenc or babel.

That should have been loaded in my class file...

@UlrikeFischer Let me check

I load both

\RequirePackage[utf8]{inputenc}
\RequirePackage[german]{babel}

@BrainStone You should never load inputenc with luatex (actually, it does no harm, the package detects that and just puts out a warning that it should not be used)

@DavidCarlisle Just not load it at all or load something differently?

@BrainStone see line 407 of your gist

Ok. Fair

I used to work with pdfLaTeX for the longest time where that is needed I think

@BrainStone but the tex you show is using memoir that doesn't input inputenc does it?

@BrainStone not needed there either if using utf-8 since the release last year (for different reasons)

@DavidCarlisle I used it in 2018

@DavidCarlisle What exactly do you mean? I genuinely don't understand what you mean/want to know of me

@BrainStone that doesn't mean it was needed:-)

@DavidCarlisle Though in any case that really shouldn't matter, should it?

@BrainStone I mean with luatex you should never use inputenc. with pdftex you may need to use it to specify an encoding diferent to utf-8 or if your latex is older than 2018

@DavidCarlisle Ok. Will remove it then

And it didn't make a difference either ;)

@BrainStone no that was just a comment, but have you compared package versions with the miktex run that worked, something must be different

As that issue seems to be getting more complex than I thought, would it make sense to post it as a question?

@DavidCarlisle I don't even know at what packages to look

I have the full CTAN suit

Both on Linux/Texlive and Windows/MikTeX

@BrainStone just put both log files side by side and step through to find the difference, no need to guess.

@BrainStone not if the question is so confused like your description here. Why can't you produce a real tex-file showing the problem along with the log-file of the compilation(s)? Why do you invent code instead?

@UlrikeFischer "invent code" what do you mean with that?

And I can try to get a minimal file that exhibits that issue

@BrainStone you shared a link to a tex file which didn't produce any of the log files you showed. That doesn't make sense.

Oh. Now I know what you mean. I had to change the class because it's a private class. Let me try to get something that matches

My apologies

x = 19
N = 6
for k in range (1,N):
    i = x
    if i % 2 == 0:
        i=i/2
        print i
        i=i+k
    else: i=3*i+1
    i=i+k
    print i

SAGE!

Could someone correct that? I am to dump. I think its only a little mistake.

` i = x` is still inside the for loop

@BrainStone Oh yes, Mr. Stone. I can not good at writing programms. :(

* I am

Not a problem

@BrainStone So could you create the correct code eventually?

@cis Yyou do know that this the chat for (La)TeX, right? Not for pyhton or coding in general

@BrainStone Yes, but we do SageTeX at all.

@UlrikeFischer I think I managed to nail it down: gist.github.com/BrainStone/684f6fb9155716175b42621b42a51472

It appears that the space in the file name is causing the issues

So maybe a bug luatex 3.1? As with 3.00 (what MikTeX uses) it works fine

Anyways, I need to go home. Will be back in 30 minutes

@BrainStone Probably you need the new latex which isn't in MiKTeX yet. Or you do what sensible people do and avoid spaces. But I can't test now, I'm out.

I think, this should be correct. But it is not....

x = 19
for k in range (0,33):
    if x % 2 ==0:
        x=x/2
        print x
        x=x+k
    else:
        x=3*x+1
        print x
        x=x+k

Ok, this is correct. Thx.

x = 19
for k in range (0,33):
    if x % 2 ==0:
        x=x/2
        print x
    else:
        x=3*x+1
        print x
x=x+k

@BrainStone 3.1? luatex is at version 1.10 (or 1.11 in development branches)

@cis you should have coded in in tex,

@UlrikeFischer I mean it works in MikTex, but not in texlive (that one being more up to date)

@DavidCarlisle My bad. I was referring to the luaoftload version. But I don't think that is it

Everything you wanted to know about Brexit but didn't ask.

@DavidCarlisle does this compile (twice) for you if the file name of the document contains a space e.g. test test.tex?

\documentclass{article}
\begin{document}

\end{document}

@UlrikeFischer doesn't work the first time

@UlrikeFischer hmph it used to work, what did we break...

@DavidCarlisle there is something wrong with the input of the aux-file. And on my system it then inputs a test.tex (as I have all sort of test files in the folder)

@UlrikeFischer yes just tracing now..

@DavidCarlisle yes. We should get a penny for every error related to it. We would be rich by now.

@UlrikeFischer OK found it

@DavidCarlisle where is it?

@UlrikeFischer here:

\makeatletter
\long\def \IfFileExists@#1#2#3{%
  \openin\@inputcheck\quote@name{#1} %
  \ifeof\@inputcheck
    \ifx\input@path\@undefined
      \def\reserved@a{#3}%
    \else
      \def\reserved@a{\@iffileonpath{#1}{#2}{#3}}%
    \fi
  \else
    \closein\@inputcheck
    \edef\@filef@und{\quote@name{#1} }%
    \def\reserved@a{#2}%
  \fi
  \reserved@a}
\makeatother
\documentclass{article}

\begin{document}

\end{document}

@UlrikeFischer replacing "#1" by \quote@name{#1}

@DavidCarlisle patch level 3 ;-)

I was just wondering about that.... I would guess so

@UlrikeFischer I wonder if l3build can cope, let's see....

@DavidCarlisle I saw another message in d.c.t.t. about the begin-document error too. There wasn't any details, but it could be the same problem.

@DavidCarlisle you mean with a test file?

@UlrikeFischer fails I think ...exlive/2019/texmf-dist/scripts/l3build/l3build-check.lua:90: ./build/test/tlb filename space.log: No such file or directory (@JosephWright) that is your document as tlb filename space.lvt

Looks like you found the issue?

Btw sorry for the space in the file name. But nothing I can do

Company standards and such

@BrainStone yes, once you mentioned the space it was easy to reproduce.

@UlrikeFischer it's same issue?

@BrainStone any standard that says anything other than "don't put spaces in filenames" should be ripped up!

@DavidCarlisle yes, at least it looks like it.

@DavidCarlisle the guy at d.c.t.t just confirmed (clearly quite surprised) that his problem was due to spaces too.

@UlrikeFischer I suspect the second fix I suggested is better, or perhaps both (in case a name gets to IfFileExist@ somehow by an unauthorised route that hasn't already normalised it

@DavidCarlisle I don't really have an opinion about it. Too much quotes and unquotes around and too late now ;-)

@UlrikeFischer yes I think best if I open open a public gh issue with some possible fixes and see if we get any comments and/or tests, I'll see what Frank says in the morning