Does anybody here use a spell checker from within emacs to check LaTeX files? If so, what do you use? (The link is just for fun.)

@HaraldHanche-Olsen boo. :)

@HaraldHanche-Olsen: on a serious note, I use hunspell.

@PauloCereda That looks like a good option from my initial look too. Do you use it as a standalone program?

@HaraldHanche-Olsen Yes, from the command line. :) I really enjoy the way the program works.

Surely there might be an emacs wrapper.

@PauloCereda Apparently, one can make ispell use it. Still looking for the documentation.

The hunspell home page claims that hunspell is built into macOS. But I don't see the hunspell command. There is a macports package, though.

@PauloCereda Ah, it seems that the ispell command in emacs can use hunspell as a backend. I'll try it out.

@HaraldHanche-Olsen Oh I never used in macOS... :(

@HaraldHanche-Olsen I always get confused with ispell, aspell and hunspell. :)

@PauloCereda Development of aspell seems to have stalled. It is what I used to use, a long time ago. But hunspell is still going strong. Regarding ispell, it seems to be both a standalone program and an emacs thingie that can talk to all three backends.

@HaraldHanche-Olsen Ah that makes sense. I think those checkers share (at some point) dictionaries. It also goes down to how a distro provides such packages. I remember having trouble in finding how OpenSUSE made hunspell available (no problem finding that information on Fedora).

Of course, I never make speiling miss steaks, but now I need to check a 450 page proceedings volume written by others. It seems I find more misspelled words every time I have a look, so I think a more systematic approach is needed.

@HaraldHanche-Olsen :D

@HaraldHanche-Olsen 450 pages, of course @PauloCereda still on page 1 so easy to check by eye..

@DavidCarlisle oi

@DavidCarlisle having a blank page also helps you are mean

@DavidCarlisle ooh

That's new :)

@PauloCereda Standard trick in martial arts: Deflect the opponent's attack.

@PauloCereda defensive tactic to stop @HaraldHanche-Olsen posting a fifty deep deja vu link

@DavidCarlisle We haven't got that deep yet. Still working on it.

Apppparently there's supper. I am an idiot.

:D

@DavidCarlisle LaTeX2e <2018-04-01> patch level 2 (with normal miktex).

@UlrikeFischer thanks, so let the fun begin:-)

@DavidCarlisle for all semiotic mappings of fun. :)

PRINT SHOP

\documentclass{article}
\RequirePackage{fontspec,unicode-math}
\setmainfont{STIX Two Text}
\setmathfont{STIX Two Math}
\begin{document}
$\gamma_{e}\ne\gamma_{\mathrm{e}}$
\end{document}

Why the different placement of the index?

Maybe I'd better ask a question on the main site.

@PauloCereda Before supper there's the glass of wine on the train.

@egreg ooh

@HaraldHanche-Olsen well the placement is governed by the font in the first case and by *tex in the second, (do you get different results in xetex/luatex?) (also try \symbrm so it uses the roman from the math font

@DavidCarlisle According to @egreg (comment on the main site), the kern is added in XeTeX. He also suggested \symrm. I was about to try that for myself.

@HaraldHanche-Olsen It is font dependent anyway: no kern is added with Latin Modern Math or Asana Math.

@HaraldHanche-Olsen That's my output with tl18: (normal/mathrm/symrm)

and this with tl17:

@UlrikeFischer Interesting. Same font in both cases?

The fonts are identical and also the fontloader. But unicode-math is different.

@HaraldHanche-Olsen ^^^

@UlrikeFischer Urgh. I'll definitely not install TL2018 until my current project is completed.

@UlrikeFischer It seems hard to blame unicode-math, though, doesn't it? luatex seems a more blameworthy target.

@HaraldHanche-Olsen It is not about blame but about how to find the source of the change. But luatex is different naturally too, so it could also be an engine change. I just tried with Tl17 and a luatex 1.06 and the kern is there, so if there was an engine change it is from 1.06 to 1.07.

@UlrikeFischer Okay, thanks, that is useful information. I just signed up to the luatex list, and will ask there as @egreg suggested on the main site. (I did use the word “blame” somewhat loosely there.)

This has made me wonder if it is safe to say \AfterEndPreamble{\let\mathrm=\symrm} in my class file. (Using that hook just in case something changes either in a BeginDocument hook.) What is likely to go wrong?

@HaraldHanche-Olsen ^^

I have this on a tee. :)

@PauloCereda :) Well, only one paper of this proceedings volume uses \mathrm a lot (the one I am currently checking). So looking for breakages seems like a manageable task. But really, so long as the argument consists of Latin letters only, it ought to be safe. And that seems to be the case here.

@HaraldHanche-Olsen phew!

@HaraldHanche-Olsen I think unicode-math has an option that does something similar for compatibility, but if you do that you break any use (multi letter identifiers) which really need \mathrm which was why \symbrm was introduced in the first place,

@HaraldHanche-Olsen if the argument is more than one letter use \mathrm not \symrm

@DavidCarlisle good point, but in this particular case (stix text and math fonts), there does not seem to be any visible difference. The usage that triggered all this was $\sigma_{\mathrm{ess}}$ for the essential spectrum.

@HaraldHanche-Olsen in general the text font will have interletter kerns and ligatures not in the math one, certainly if you compare \mathbf and \symbf you will get big differences

@DavidCarlisle all right; I shall be careful. I'll mark that code change as potentially deadly. Committing horrors to get this volume pushed ut the door.

Wondering if I could hack it by creating a zero width, invisible symbol from the math font and stick it in front of \mathrm{ess}. Perhaps that would trigger the necessary kern. But if I go that route, maybe it's just as easy to just edit in a kern in all the relevant locations. (Or make a macro to replace \sigma_{\mathrm{ess}}`.)

@PauloCereda @DavidCarlisle I assume hunspell ignores spaces between words...

@AlanMunn DubrauchsteineRechtschreibprüfungfürDeutsch.EnglischeRechtschreibungistvieleinfa‌​cher.

@DavidCarlisle Ich sehe, was du getan hast.

@HaraldHanche-Olsen Imho you will need a mathkern table see Math kerning in super- and subscripts in the luatex documentation. But I don't know any example.

@AlanMunn hm

@UlrikeFischer As an emergency workaround, I just created some macros for those two cases where this is an actual problem. There is actually a quite visible difference between \mathrm{disc} and \symrm{disc}, as it turns out.

\newcommand\spdisc{\sigma_{\mkern-3mu \mathrm{disc}}}
\newcommand\spess{\sigma_{\mkern-3mu \mathrm{ess}}}

@AlanMunn Just like FORTRAN!

@HaraldHanche-Olsen :)

Here are all your favourite (and not so favourite) languages with funny quotes about them from famous people to match. www-users.cs.york.ac.uk/susan/cyc/p/prog.htm

Can someone please help to reopen tex.stackexchange.com/questions/418308/… ? Only one vote is missing.

@samcarter Done

@PauloCereda Thank you!

@samcarter <3

@PauloCereda Was your visit at the printing shop sucessful?

@samcarter I am still waiting for the result. :) It takes a while.

@PauloCereda Ohh! So many pages to print?

@samcarter yes, and 6 copies each.

5 paperbacks and one hardcover.

@PauloCereda So you killed a few trees :)

@samcarter sadly. :)

@PauloCereda I bet the trees don't mind as it is for a good cause!

@samcarter :D

@DavidCarlisle oh no

@DavidCarlisle You asked for it. ↑↑

@PauloCereda I need to get them in while I can

@DavidCarlisle <3

@AlanMunn ^^

cbmountaintheatre.org/2018-golden-marmots-award-show.html

@marmot ooh marmot awards

@PauloCereda I didn't grow up watching Sesame Street, but this is very cool.

@AlanMunn and @PauloCereda -- i didn't grow up watching sesame street either, but when i was in my early 30s, i spent three months in hospital, and sesame street was one of the few things that kept me from going out of my mind with boredom and frustration. they do wonderful work -- it's almost magic!

@AlanMunn it is!

@barbarabeeton How nice, Barbara!

Re my earlier tribulations with subscripts (see two–four screenfuls above): I found a workaround: With \Umathchardef\zwspace="0"0"200B I can say $\gamma_{\zwspace\mathrm{e}}$ and get the subscript placed correctly. This works presumably because U+200B ZERO WIDTH SPACE has, well, width zero, and it exists in the STIX Two Math font.

They have the Kickstarter campaign: kickstarter.com/projects/sesamestreet/…

@HaraldHanche-Olsen -- sneaky! this should get wider coverage. just have to figure out where it's most likely to be seen and remembered ...

@barbarabeeton But it's not perfect: If the next character is tall, the whole subscript could wind up too far to the left, since that character is not taken into account at all. Good enough for my present needs, though.

@HaraldHanche-Olsen I think luatex's treatment of kerns in math mode is unfortunate, Hans claims all the fonts are wrong (and that's not a totally unreasonable view:-), but that doesn't help the end user

@HaraldHanche-Olsen have you tried setting \mathitalicsmode and \mathscriptboxmode ?

@DavidCarlisle No, since I haven't heard about them until now.

@DavidCarlisle Ah, section 6.8 and 6.9 in the luatex manual. I'll have a look. But right now I am packing up to go home.

@DavidCarlisle -- is it assumed (by authors, at least) that amsmath should work the same in luatex as in pdftex et al? i know that it will be necessary to make amsmath get around the unfortunate absence of a parameter in the otf math table, but it's not reasonable to make it accommodate an entirely different way of handling kerning. seems to me that might affect the performance adversely and possibly cause other problems that could make it unsuitable for ams production. not a satisfactory situation.

@DavidCarlisle I tried \mathitalicsmode and it didn't do anything (but it doesn't do anything in the docu either so perhaps this is intended ;-)

@JohnDorian you might be better off making \inv take an argument that way if you change how you format the inverse in the future, you won't have to change the body of the text.

@UlrikeFischer Thank you! I thought it was just me getting too old to see the subtle differences there.

@HaraldHanche-Olsen Btw try $\gamma_{e^2} \gamma_{ee} \gamma_{e\hphantom{\strut}} \gamma_{e} $

@JohnDorian if it helps you and doesn't conflict there is no reason why you shouldn't use it.

@UlrikeFischer The last one doesn't make any sense. And I can't reproduce it.

@UlrikeFischer It seems a bug: when the subscript field has just one atom, the kern is not inserted.

@egreg Yes that's what I thought too.

@egreg Is that on TL2018?

@HaraldHanche-Olsen Yes.

@HaraldHanche-Olsen Yes. If I do $\gamma_{\kern0pte}\ne\gamma_{\kern0pt\mathrm{e}}$, the negative kerns are applied.

\mathon
\TU/STIXTwoMath(1)/m/n/10 𝛾
\kern-1.5 (font)
\hbox(3.6254+0.09818)x3.85008, shifted 2.09998, direction TLT
.\kern0.0
.\TU/STIXTwoMath(1)/m/n/7.01236 𝑒
\glue(\thickmuskip) 2.77771 plus 2.77771
\TU/STIXTwoMath(1)/m/n/10 ≠
\penalty 500
\glue(\thickmuskip) 2.77771 plus 2.77771
\TU/STIXTwoMath(1)/m/n/10 𝛾
\kern-1.5 (font)
\hbox(3.401+0.07013)x3.65373, shifted 2.09998, direction TLT
.\kern0.0
.\TU/STIXTwoText(1)/m/n/7.01236 e
\mathoff

:43923896 I guess it's the indication the kern is inserted because of font information.

Never mind, my eyes failed me. It's in both places.

Ahoi TeXians! I have been developing a small tool (and library) that makes LaTeX logs more readable. I'm sure I missed plenty of things, so your feedback would be much appreciated: github.com/reitzig/texlogparser

@Raphael With pdflatex -file-line-error -interaction=nonstopmode test | texlogparser I get no output at all.

@egreg Works without a hitch here? Huh. Which OS are you running?

@Raphael ruby 2.0.0p648 (2015-12-16 revision 53162) [universal.x86_64-darwin16]

(Of course it couldn't just work for anybody but me.)

Uh. Ancient Ruby on a Mac?

Never tested with anything below 2.3, sorry.

@Raphael Sorry, apparently it took the wrong ruby. Let me try again

@Raphael I installed ruby 2.5.0 with homebrew, but apparently the shell still looks for the older one

Strange:

Ah, I remember. Getting it to switch was somewhat painful.

mar2018 > env ruby --version
ruby 2.5.1p57 (2018-03-29 revision 63029) [x86_64-darwin16]
mar2018 > ruby --version
ruby 2.0.0p648 (2015-12-16 revision 53162) [universal.x86_64-darwin16]

Yup. X-D

@barbarabeeton the genfrac math params are not an issue, that's handled already. but you can't ask that math kerning is handled the same was as in classic tex as the math layout if using opentype fonts is almost entirely different to the math layout using tfm fonts, there are some similarities because Tex was used as part of the model for opentype math, but basically it is a separate math layout engine

@Raphael The first line of texlogparser seems suspicious:

texlogparser has this shebang: #!/usr/bin/env ruby So...

#!/usr/local/opt/ruby/bin/ruby

Hah! What on earth did gem build do there?

@Raphael I don't know

Makes sense, though: gem hard-codes its own Ruby. Try gem --version vs env gem --version?

If the latter is newer, gem uninstall tex_log_parser and then env gem install tex_log_parser.

Or maybe just env texlogparser to try it out quickly.

@Raphael Still get the same shebang

._.

@Raphael Same result: no output

I don't have a Mac handy so I can't really help, sorry. :/

No output is weird, I'd expect an error at least.

@Raphael I get output if I add -d at the end

@Raphael But I guess it shouldn't be needed

@DavidCarlisle -- so that implies that amsmath (unless modified) can't work with opentype fonts? is that one of the reasons we haven't had the results we were hoping for with stix 2?

@egreg Nope, but what do you get?

(It seems that rbenv may be helpful with the Ruby on macOS mess.)

mar2018 > pdflatex -file-line-error  -interaction=nonstopmode gray | texlogparser -d

Line 1: 'This is pdfTeX, Version 3.14159265-2.6-1.40.18 (TeX Live 2017) (preloaded format=pdflatex)'
- No pattern matches

Line 2: 'restricted \write18 enabled.'
- No pattern matches

Line 3: 'entering extended mode'
- No pattern matches

Line 4: '(./gray.tex'
- No pattern matches
- Entered source file: './gray.tex'
[...more lines...]

@Raphael ^^^^^

That looks good. What's the ` ... | tail -20`?

Line 8: 'Document Class: article 2014/09/29 v1.4h Standard LaTeX document class'
- No pattern matches

Line 9: '(/usr/local/texlive/2017/texmf-dist/tex/latex/base/size10.clo)) (/usr/local/texlive/2017/texmf-dist/tex/latex/booktabs/booktabs.sty) (./gray.aux) [1{/usr/local/texlive/2017/texmf-var/fonts/map/pdftex/updmap/pdftex.map}] (./gray.aux) )</usr/local/texlive/2017/texmf-dist/fonts/type1/public/amsfonts/cm/cmbx7.pfb></usr/local/texlive/2017/texmf-dist/fonts/type1/public/amsfonts/cm/cmr10.pfb></usr/local/texlive/2017/texmf-dist/fonts/type1/public/amsfonts/cm/cmr7.pfb>'

Also good. It just doesn't output the result. Curious.

@barbarabeeton well it works because amsmath has very little code at the level of math atom positioning, alignments etc all work as expected because they are handled by the tex box code rather than font shaping. But inter letter positioning sub/superscript positioning subscript sizes etc are basically different (and under specified) there are differences in layout between luatex xetex word and firefox all implementing the same opentype math spec because the spec is er vague in many places

@Raphael Always empty output

@Raphael With -f json I get [<newline><newline>]

Ah, so output works, technically. It "only" loses all the messages. Well, if there are any.

Can you try with one of these, please?

Heh, looking at what you pasted, if there wasn't something extra special in log lines 5-7, maybe there just aren't any messages? In that case, I'm perfectly content with the output. :D

# TODO: print summary? <-- from the script. As if I'd known. :'D

@Raphael Oh, well! No info at all? Not really what I'd expect.

@Raphael Yes, if I plant an error the output is there.

Phew! :D

@Raphael Some more customizations should be needed, I guess.

Well, what would you like to see? For me, it was always about seeing errors and warnings, and filtering out all the rest.

@Raphael For instance, messages about underfull or overfull boxes and the last line (how many pages).

(I do note that it doesn't pick up on the closing paren of (./gray.tex, which seems to be in an awkward position. So if you could forward me your source file, that would be kind. :))

@egreg Under-/Overfull boxes are implemented. There aren't any, at least not in the fraction you posted?

I'll expand the ToDo: print as summary how many messages there were, and which files were written (here, PDF and log).

@Raphael There was one. If I use texlogparser -d on the following file, I get still Files that did not close: ["./test.tex"]

\documentclass{article}
\begin{document}
abc
\end{document}

@Raphael you need to show file open and close the (....) unless you parse that and report the filenames elsewhere otherwise when there are errors you don't know which file they refer to

@Raphael If I change abc into \hbox to100pt{abc}, there is an underfull box message in the log , but it doesn't appear with texlogparser

@Raphael The log file for the “non underfull” version

This is pdfTeX, Version 3.14159265-2.6-1.40.18 (TeX Live 2017) (preloaded format=pdflatex 2018.2.15)  10 APR 2018 23:22
entering extended mode
 restricted \write18 enabled.
 %&-line parsing enabled.
**test
(./test.tex
LaTeX2e <2017-04-15>
Babel <3.18> and hyphenation patterns for 84 language(s) loaded.
(/usr/local/texlive/2017/texmf-dist/tex/latex/base/article.cls
Document Class: article 2014/09/29 v1.4h Standard LaTeX document class
(/usr/local/texlive/2017/texmf-dist/tex/latex/base/size10.clo

@Raphael Anyway, thanks for contributing

@DavidCarlisle Sure. The script does that, or tries to, anyway.

You'll get output like this:

Extraction is not perfect, certainly.

@egreg I don't see the underfull box message there myself?

@Raphael If you want the one with the underfull, here it is

This is pdfTeX, Version 3.14159265-2.6-1.40.18 (TeX Live 2017) (preloaded format=pdflatex 2018.2.15)  10 APR 2018 23:26
entering extended mode
 restricted \write18 enabled.
 %&-line parsing enabled.
**test
(./test.tex
LaTeX2e <2017-04-15>
Babel <3.18> and hyphenation patterns for 84 language(s) loaded.
(/usr/local/texlive/2017/texmf-dist/tex/latex/base/article.cls
Document Class: article 2014/09/29 v1.4h Standard LaTeX document class
(/usr/local/texlive/2017/texmf-dist/tex/latex/base/size10.clo

@egreg Can't reproduce. I guess our logs look different?

@Nasser Only text, of course

@egreg thanks. So is there different command to set math font similar to the above using fontspec?

@Nasser You need unicode-math (and a Unicode math font)

@Nasser two main packages mathspec or unicode-math (use unicode-math if you are using an OpenType Math font)

I see. Ok, thanks. Will look now into mathspec then.

@Nasser if you are using a proper opentype math font use unicode-math mathspec is for fonts without a math table

@egreg Perfect, thanks! github.com/reitzig/texlogparser/issues/1

(Sloppy mistake on my part, I remember thinking about that. "Is it always at lines XX-YY? Ahh, probably.". Pfft.)

@DavidCarlisle I do not understand what you just said actually mean, but I will try to find out :) . Using fonts in Latex is something I never understood and I do not think I will ever do. Just trying to find nicer font to use for math and text. But many places give different commands, and hard to know what to use and where. someone should write an article: using fonts in latex for dummies.

@Raphael if I try

\documentclass{article}

\def\zzzz{\zzz}
\def\zzz{\zz}

\begin{document}


aaa\\


\zzzz



\end{document}

then pdflatex abc3 | texlogparser just hangs, I seem to need pdflatex -interaction=nonstopmode abc3 | texlogparser

but then the output is

./abc3.tex:12: ERROR
! Undefined control sequence.
\zzz ->\zz

\zzzz

@egreg The underfull hbox warning was easy to fix. The missed ) is harder.

{/usr/local/texlive/2017/texmf-var/fonts/map/pdftex/updmap/pdftex.map}] (./test.aux) )

so it loses the underfull box error even though it is a top level paragraph one with "lines" not an explicit hbox one as in egreg's example

Underfull \hbox (badness 10000) in paragraph at lines 9--10

Who does that?!

@Raphael who does what?

@DavidCarlisle Sure without nonstopmode, pdflatex will stop and wait for interaction.

@DavidCarlisle Outputs log lines like that one.

@Raphael sorry but that makes it unusable if it stops for interaction but shows no prompt or message.

@Raphael I do not understand, you mean you who outputs the underfull hbox warning? Hopefully all tex users.

@DavidCarlisle Yup, it's not designed for that. I guess if you want to interact directly with the interpreter, you're also fine with it's unfiltered output. My goal is to improve "compiler" usage. So, yea, won't fix.

I meant: Who on earth things that a log format where (file and ) can be scattered all over the place is a good idea?

@Raphael oh ignore the comment I deleted:-)

@Raphael I get in full:

$ pdflatex -interaction=nonstopmode abc3 | texlogparser
./abc3.tex:12: ERROR
! Undefined control sequence.
\zzz ->\zz

\zzzz

Hmm. I think I see the problem. texmf (?) is running (on first runs?) and keeps injecting its own output into that of pdflatex, muddling up the output.

Running on the .log which pdflatex writes is more robust, or on the output of a second run.

@DavidCarlisle Hmjea, it seems not all Under/Overfull messages end with [] after all.

@Raphael well you get [] if the \showboxdepth is low enough not to show the box contents, in that case you get

Underfull \hbox (badness 10000) in paragraph at lines 9--10

 []

Oh dear lord...

So... sometimes the end it [], sometimes an empty line?

@Raphael if you increase \showboxdept you do not get [] and instead you see the box contents:

In that case you just posted, the empty line is part of the message?

Underfull \hbox (badness 10000) in paragraph at lines 9--10


\hbox(0.0+0.0)x345.0
.\glue(\rightskip) 0.0

And no definitive delimiter of the message?

@Raphael yes

@Raphael yes

Too bad then. Have to parse it in some way -- empty line it is, and stuff will have to be cut off sometimes. :/

@Raphael although to be honest I am not your target audience as more often than not I add \tracingall to the preamble (and get 3997 lines of log from that example file:-)

@DavidCarlisle Oh jeez. X-D Let's table that for another day. :)

@egreg and @DavidCarlisle, thanks for your feedback! Even though you may not be the precise target audience, you already helped improving the tool!

It's a litte depressing how easy I made it for you, but well.

The poking holes into it, I mean.

@Raphael over the last 30 years I have probably spent much longer looking at \tracingall log output than I have the typeset output of tex...

That one I'll have to think about. Treating any parens outside of actual messages as file change operations is probably worse than missing one on occasion.

@DavidCarlisle I have never looked at \tracingall output and I already feel bad for you. My programmer soul can barely handle the normal logs.

@Raphael since you are anyway saying to use --interaction you could tell people to use --file-line-error then at least for errors the file name and line number are reported in a half way sensible format

so, does this mean that the text font and the math fonts can be selected to be different? for example, I see nice font called ITC New Baskerville. But when I go to buy it, I only see text samples: myfonts.com/fonts/itc/new-baskerville no math. But then what about math font that will go with it? Is one supposed to select both text font and then math font separately? and try to see if they look good together?

@Nasser there are rather few math fonts the tex gyre project, cambria math,asana math, stix math stix two math, and maybe a couple of others. oh and some commercial ones lucida, minion pro math, ..

@DavidCarlisle what do you recommend/like for text font and math font?

minion pro math, for math. How does one know what to pick for text then? This is getting confusing.

@Nasser I have used computer modern for the overwhelming majority of my tex documents (and currently i don't actually use tex at all for any documents)

@DavidCarlisle you do not use Latex for your documents?? OMG! Please do not say you use MS word??

@DavidCarlisle And which font do you use for your Word documents? ;-)

@Nasser it's like asking someone what colour should they paint a wall, it depends what colour you want

@Nasser I have never made any document in any wysyswig word processor, no we type xml in emacs and convert to html (previously we also converted to pdf via 3b2 but we just stopped that and just do html now)

@marmot comic sans (@egreg recommended it for Word use)

@DavidCarlisle why not just use tex4ht to convert latex to HTML? writing in XML must be really harder than latex. specially if there is math involved.

@Nasser that wouldn't be a good idea:-)

@DavidCarlisle I use latex for everything and convert to HTML all day using tex4ht. It works really well.

@Nasser from the same xml file we generate, fortran library documentation, C libray documentation, matlab toolbox documentation pydoc/sphinx documentation, the C source code for the matlab interface, the python code for the python interface the C source code for the C interface, C headers for the fortran library, VB and VB.net interfaces, C# interfaces and a few other things. Having validated xml input is essential

@Nasser I just did a quick count there are something like 80 thousand separate math expressions in the sources.

@DavidCarlisle I do say that, right in the README! :D As you say, it makes for more robust error reporting in case the parens business doesn't pan out.

@Raphael I never read all the readme of course:-)

github.com/reitzig/texlogparser#recommendations <-- there you go :P

@DavidCarlisle you must be really good in XML. I can't imagine writing math in XML. I also use Scientific word, which is a Latex front end and it generate latex for me. Makes it really easy to type complicated math expressions quickly and easily. But I see that XML fits what you do best. I only need Latex and tex4ht for my purposes.

Bed time for me. The bugfix is tested, pushed, and released. If you're still interested, you can gem update and see some more hbox warnings detected.

Thanks again!

@Nasser my job title is "Group Leader, XML Technologies Group" so hopefully I know something about XML:-)

@Raphael oh yes :-)

@Raphael Sorry, again empty output notwithstanding the underfull box.

> texlogparser --version
texlogparser 1.0.1