@FaheemMitha I do the following: click on a featured comment (with more than a day long) -> pick another day than today (for example, yesterday) -> that will generate a link with the format .../AAAA/MM/DD, e.g., https://chat.stackexchange.com/transcript/41/2018/9/20. You are free to find a particular day knowing the year, month and day and changing it into the link :)

@manooooh Thank you for the tip.

Does the 41 correspond to the room?

@FaheemMitha yes!

We can also filter by room haha, that's awesome!

@manooooh Ok, thank you.

Few hours to see Atiyah's proof. Will he be right?

If there are something interesting please ping me! Here it will be 4.45 a.m. when he exposes his proof and I will not be able to see it until after several hours

@samcarter TeXStudio uses a TikZduck now, see my edit here: tex.stackexchange.com/questions/332307/…!

@CarLaTeX :) they even change depending of the time of the year github.com/texstudio-org/texstudio/pull/1/commits/…

@samcarter :):):)

oiy, tex.stackexchange.com/q/452215/3929 can actually be converted as the sources are publicly available at the arXiv website. So in this case it IS possible, but I cannot add an answer when the question is closed

@daleif Not really a conversion.

@egreg true, but there is an available .tex for this particular PDF, the OP might just not realize it.

Well, leave it then, the OP will just have to read my comment instead and perhaps phrase their question a bit narrower next time

Should the following be closed as off-topic? tex.stackexchange.com/questions/451450/…

@Skillmon I have voted to close (seems to solely revolve around a company's services).

@TeXnician seems reasonable.

@samcarter These guys are cruel. Not a single koala.

@marmot The logical explanation for the missing coala is that this change predates the tikzlings. Today they surly would include the whole tikz fauna!

@samcarter Let's keep our fingers crossed. ;-)

@marmot can koalas cross their fingers?

@DavidCarlisle I think so, but I also said : "Let's keep our fingers crossed" So far there are not too many koalas in this chat, sadly.

@marmot can marmots cross their fingers?

@DavidCarlisle Of course! (You may call them claws, though. ;-)

@DavidCarlisle It's your (and the rest of your countries) fault. If we would have this discussion in German, the gesture would be much easier: de.wikipedia.org/wiki/Daumen_dr%C3%BCcken and koalas even have two thumbs per hand, so they can do this two times as effective!

@samcarter but being a mathematician I know that 2 times 0 is 0

@DavidCarlisle Can you also flap your ears? i.gifer.com/iZO.gif

@DavidCarlisle That is a true statement, but I don't see any relation to the koala problem :)

@samcarter your statement the line above that it's twice as effective

@DavidCarlisle I simplified the factor problem a bit:

\directlua{
    local cbl=luatexbase.callback_descriptions('define_font')
    if(cbl[1]\string~=nil) then
      original_fontloader=luatexbase.remove_from_callback('define_font',cbl[1])
    end
function latexDefineFont(n,s,i)
  for ii,vv in font.each() do end %comment this
  return original_fontloader(n,s,i)
end

luatexbase.add_to_callback('define_font',latexDefineFont,"latexDefineFont")
}

\documentclass{article}
\begin{document}
\section{Test of math accents}

$x$

\section{Test of roman text in math}

@DavidCarlisle tex.stackexchange.com/questions/126903/…

@UlrikeFischer ooh google for error messages ? :-)

@UlrikeFischer so it's user error in the callback code? and just lucky it seemed to work before?

@DavidCarlisle not really, after I saw that font.each() is the problem I remembered that Philipp had written at some time about font access. I only saw the error message when I found the question.

@DavidCarlisle yes looks so.

@UlrikeFischer although perhaps the font loader could be a bit defensive and return an error if that is nil rather than letting the engine segfault and leave Segmentation fault (core dumped)

@DavidCarlisle The segfault is new. With 1.07 I get simply an error in the terminal. It crashes with 1.09. But we probably need a simpler example for a bug report.

Thought people here might find this humorous. Yesterday when searching for an answer by a particular member (as I recalled that I had previously found that useful) I typed in the search box user:4427 and also 'is:answer` before realizing how redundant that was... :-)

@PeterGrill I agree it's never worth searching for that person's answers.

@DavidCarlisle: -) Don't you mean never worth searching questions by that person!

@PeterGrill no his questions are fine, it's his answers that I have doubts about:-)

@DavidCarlisle :-) :-) :-)

@PeterGrill He's referring to user 4427-47*71, just forgot to subtract

@egreg nothing wrong with those answers ^^^

@DavidCarlisle It would be a good occasion for merging xr and xcite

@egreg do I take that as you volunteering? (and xr-hyper)

@DavidCarlisle I can lend you the code.

@egreg running test suite for the fontdef change at present...

@egreg :-) :-) Had to work that out!!

@DavidCarlisle did you adapted the fontloader code?

@UlrikeFischer no not sure I understood (and got distracted) meanwhile since I updated l3build I can't run the test suite (@JosephWright)

No pages of output.
Transcript written on ../build/unpacked/unpack.log.
Running checks on
  git-0043 (1/352)
warning: kpathsea: configuration file texmf.cnf not found in these directories: .:/usr/local/texlive/2018/texmf/web2c.

kpathsea: Running mktexfmt latex.fmt
warning: kpathsea: configuration file texmf.cnf not found in these directories: .:/usr/local/texlive/2018/texmf/web2c.
/usr/local/texlive/2018/bin/x86_64-cygwin/mktexfmt: kpsewhich -var-value=TEXMFROOT failed, aborting early.
BEGIN failed--compilation aborted at /usr/local/texlive/2018/bin/x86_64-cygwin/mktexfmt line 25.

@DavidCarlisle Hmm, same issue as reported in an email to me. I don't see it on Windows or my Linux VM

@DavidCarlisle Something somewhere is up when setting TEXMFCNF as an environmental variable

@DavidCarlisle I guess a cut down test would be handy to try to track down whey it's happening: from what the mail I got said, kpse.var_value("TEXMFCNF") returns the right thing ...

@DavidCarlisle One option is only to set this at all if requested, so an opt-in probably used just by @UlrikeFischer. But I don't see what is fundamentally wrong ...

if it helps, outside l3build, the enviornment variable isn't set, and kpse variable is:

$ kpsexpand '$TEXMFCNF'
{/usr/local/texlive/2018/bin/x86_64-cygwin,/usr/local/texlive/2018/bin/x86_64-cygwin/share/texmf-local/web2c,/usr/local/texlive/2018/bin/x86_64-cygwin/share/texmf-dist/web2c,/usr/local/texlive/2018/bin/x86_64-cygwin/share/texmf/web2c,/usr/local/texlive/2018/bin/x86_64-cygwin/texmf-local/web2c,/usr/local/texlive/2018/bin/x86_64-cygwin/texmf-dist/web2c,/usr/local/texlive/2018/bin/x86_64-cygwin/texmf/web2c,/usr/local/texlive/2018/bin,/usr/local/texlive/2018/bin/share/texmf-local/web2c,/usr/local/texlive/2018/bin/share/texmf-dist/web2c,/usr/local/texlive/2018/bin/share/t…

@JosephWright did you add a semicolon/colon at the end?

@UlrikeFischer No, but I shouldn't need to

@UlrikeFischer Note this works on Windows and on my Ubuntu VM

@JosephWright but how should kpathsea then find the main texmf.cnf?

Well, Atiyah has disappointed us once again...

@UlrikeFischer Because I've not removed anything from TEXMFCNF

@JosephWright well if I do in a cmd window set texmfcnf=path/to/a/specialtexmfcnf-folder/ then texlive breaks. I have to do set texmfcnf=path/to/a/specialtexmfcnf-folder/; with the semicolon at the end.

@UlrikeFischer I'm doing the equivalent of set TEXMFCNF=.;%TEXMFCNF%, which should be fine

@UlrikeFischer Like I said, it doesn't break for me, so there must be something different about the Windows/Ubuntu installations I have, and @DavidCarlisle's Cygwin set up (and the one I've been told about by mail: also a Linux rig)

@manooooh Could you please share the link (or the source of the information) with us?

@JosephWright How do you get the %texmfcnf%?

@JosephWright and why don't you simply do texmfcnf= .; instead of trying to append the old value?

@UlrikeFischer I was just going to say that!

I have

  print ("==================")
  print (var_value("TEXMFCNF"))
  print ("^v^v==============")
  print (os_setenv .. " TEXMFCNF=." .. os_pathsep .. var_value("TEXMFCNF"))
  print ("==================")

  for i = 1, checkruns do
    errlevels[i] = run(
      testdir,
      -- No use of localdir here as the files get copied to testdir:
      -- avoids any paths in the logs
      os_setenv .. " TEXINPUTS=." .. (checksearch and os_pathsep or "")
        .. os_concat ..
      -- Avoid spurious output from (u)pTeX

and it works (texmf setting disabled of course) but the log shows

==================
{/usr/local/texlive/2018/bin/x86_64-cygwin,/usr/local/texlive/2018/bin/x86_64-cygwin/share/texmf-local/web2c,/usr/local/texlive/2018/bin/x86_64-cygwin/share/texmf-dist/web2c,/usr/local/texlive/2018/bin/x86_64-cygwin/share/texmf/web2c,/usr/local/texlive/2018/bin/x86_64-cygwin/texmf-local/web2c,/usr/local/texlive/2018/bin/x86_64-cygwin/texmf-dist/web2c,/usr/local/texlive/2018/bin/x86_64-cygwin/texmf/web2c,/usr/local/texlive/2018/bin,/usr/local/texlive/2018/bin/share/texmf-local/web2c,/usr/local/texlive/2018/bin/share/texmf-dist/web2c,/usr/local/texlive/2018/bin/share/texmf/…

the .:{xxx form looks a bit odd

@marmot yes for sure

@DavidCarlisle does it work if you use os_setenv .. " TEXMFCNF=." .. os_pathsep ?

@UlrikeFischer yes (@JosephWright)

@UlrikeFischer Because that's what one always does, for example when adding to the path. Take the existing value, recover it, prepend or append as required, set the new value

@DavidCarlisle bash shoudln't be involved: it's an environmental variable passed to kpsewhich (no bash on Windows!)

@JosephWright bash is the environment

@DavidCarlisle No

@DavidCarlisle The system is the environment: we never call an interactive shell

@DavidCarlisle But we are doing set <name>=<value>: there's no processing of the <value>, it's a string in the environment

@JosephWright hmm OK I'll have to check what os_setev is doing:-)

@DavidCarlisle Thar's export on Linux

@DavidCarlisle, @UlrikeFischer What is odd, as I've said, is that the approach as it stands works fine on my Ubuntu VM

@UlrikeFischer, @DavidCarlisle OK, I think I see what you are both getting at ... I'll adjust and cross fingers (still think there is something odd)

@JosephWright kpathsea path expansion is complicated. But as it explicitly has a method to add the existing path (see 5.3.1 Default expansion) one should use it.

@manooooh Gracias!

@JosephWright os_setenv = "export" but export is (here at least) a shell built in so it'll set the environment variables via the shell (bash in my case as far as I can see)

Hello!

I am having the same problem that I posted a few months ago (but I delete it because the solution was easy), but now I am having the same problem again...

@manooooh Maybe you should find again an easy solution and then delete it?

@marmot haha no! I am just describing my problem here, because I think I'm fool

@manooooh Is it easier or more difficult than proving the Riemann conjecture?

Shhh you deconcentrate me!

I think it is more difficult

Oh my god now it's working

The problem was with GPL Ghostcript; it didn't generate a PDF output

But now I create a ´.tex´ in a different directory than ´.../Desktop´, e.g. .../Desktop/folder/ and works for this example and with my original file

It was too weird...

Even in the log file I couldn't copy the message

I just created a new file, compiled it and now it goes. Until 5 minutes ago I could not (and I did not press any weird button, I just compiled)