@JosephWright I'm out all day so perhaps someone else would have to add a remark at the miktex github if the graphics problem persists.

@UlrikeFischer Looks like it's already been fixed

@JosephWright good :-)

Quack!

@PauloCereda Anyone for Sunday Roast?

@DavidCarlisle oh no

@DavidCarlisle I see the case changing functions have attracted attention

@JosephWright I wish to register a complaint

@JosephWright LGR :-)

@DavidCarlisle Well yes, but that's not really the issue, as I've said

@JosephWright yes, using tl_head/\@firstofone on utf8 isn't going to work true.

@DavidCarlisle Although you are right in the sense I've only covered western European fully for 8-bit: Greek, etc., really does need a Unicode engine

@JosephWright on Petra's mail, that's a new check rather than a new issue isn't it, the errata files haven't been in base for a while have they?

@DavidCarlisle We could use the existing mechanism to define some \tl_first_upper_case:n, but my concern is that this really isn't what Unicode describe, and I'm suspicious of ending up with `XxYYYYYY'

@DavidCarlisle Probably not: I guess I missed them out of the build list

Microsoft appears to be porting its Edge browser to Linux, reports ZDNet: "We on the MS Edge Dev team are fleshing out requirements to bring Edge to Linux, and we need your help with some assumptions," wrote Sean Larkin, a member of Microsoft's Edge team.... Chrome, of course, is already available for Linux, so Microsoft should be able to deliver Chromium-based Edge to Linux distributions with minimal fuss.... [I]n June Microsoft Edge developers said there are "no technical blockers to keep us from creating Linux binaries" and that it is "definitely something we'd like to do down the road".…

@DavidCarlisle ^^

@PauloCereda well given that they ship a linux kernel as part of windows, being able to run their own browser on that would make sense...

@DavidCarlisle Tracking down the issue

@DavidCarlisle bah no IE :)

@DavidCarlisle Bisecting

@JosephWright we already have at least the latex book errata on latex-project.org I'd be tempted to take them out of base and ctan and just put them on the website, but whatever Frank says, they are mostly his errors being logged:-)

@DavidCarlisle oh no

@PauloCereda edge is actually a lot better than IE and edge with a blink rendering engine is going to be very like Chrome

@DavidCarlisle So I heard. :) But I don't use any browsers other than Firefox, so... :)

@DavidCarlisle Hmm, seems to be a build script failure: I'm working on it

@DavidCarlisle Oh, I remember ...

On title case isn't the issue that you want sentence case ie just affect the first letter of the first word

@DavidCarlisle Well yes ... but ... what if you are supplied with an all-uppercase phrase? With 'first upper' that is fine most of the time, but not if the first char is one that has a 'titlecase' variant. You need to know if the rest of the phrase is going to be upper or lower case, and that can only happen if you case-change everything

@JosephWright not always, we need uppercase first letter all the time at work as we expand boiler plate strings that may or may not come out at the start of a sentence so you need to apply title case to just the first word, you don't want to change every word in the string just because it landed at the start of a sentence.

@DavidCarlisle Unicode describe titlecasing but not 'first uppercasing', I think in part as one could be supplied with poorly defined input, and because of the mixed case business: something like "LJabc" with a one-char "LJ" should look like "Ljabc", but that's no good if you start from "LJABC" (=> "LjABC")

@DavidCarlisle Yes, sure, but that's a different problem: Unicode 'titlecase' really about a single word: you need additional logic to identify 'the first word'

@JosephWright the link you gave says that the title case transform is used for sentence by applying it to the first word, so the assumption is that you can split on the first space before doing the transform

@DavidCarlisle We could alter 'mixed' case so it doesn't do anything to the input beyond the first 'character'

@DavidCarlisle This is true

@DavidCarlisle The functions are marked as experimental for a reason: Unicode stuff is hard

@DavidCarlisle Making 'mixed' case a no-op for everything except the first char would make life easier :)

@DavidCarlisle I'll make some comments in the issue

@JosephWright logically you probably need both, but I guess it is more efficient to have a flag in mixed case to only change one letter rather than wrapping it in a function that splits off a single word first

@DavidCarlisle Perhaps: I think we need a little thought/discussion on what interface works. Internally it's all doable using a flag, as that is more-or-less what happens anyway to switch to lowercasing

Meanwhile, working on the 2020 plan

I'm seeing this message in the logs of a file, but I've no idea what it means. I don't think it's an error, or even a warning.

\par  on input line 70

If the meaning is clear from the message, let me know what it means, please.

Also, shouldn't GenericWarning write to the log? It doesn't seem to be doing so here.

Are any of the syntax checkers mentioned at the end of en.wikibooks.org/wiki/LaTeX/Errors_and_Warnings#Error_messages useful?

@FaheemMitha I don't remember seeing this one. Probably some package redefines \par to do that... MWE?

@FaheemMitha It does. See: tex.stackexchange.com/a/508813/134574

@PhelypeOleinik I have a letter template. I guess I could strip it down. Should I post it here or on the site?

@PhelypeOleinik Redefines \par to do what?

@FaheemMitha Here is fine. It shouldn't be complicated

@FaheemMitha To print that message (and possibly other more useful things)

@PhelypeOleinik Oh, odd. It should show up, then.

@PhelypeOleinik Oh, I see. Not sure why it would do that, but ok.

@FaheemMitha It does not?

@FaheemMitha Packages do weird things sometimes...

@PhelypeOleinik Not as far as I can see. I'll try a small test case. Note that I'm using LuaLaTeX, though I don't see why that would make a difference.

@FaheemMitha It certainly would. My analysis in that answer is for TeX (and possibly pdfTeX and maybe XeTeX). LuaTeX has access to parts of TeX which other engines don't, so you may be able to write things on the terminal and not the log. Although the standard behaviour of \GenericWarning is to print to both

@PhelypeOleinik Ok. Trying a small test case now.

Is this a swastika? tex.stackexchange.com/questions/511145/…

@CarLaTeX I tried to find similar images on the net, but Google finds none. It looks quite suspicious.

in the morning on a tour on the etna - In the Land of Mordor where the shadows lie

@CarLaTeX @JosephWright I found en.wikipedia.org/wiki/Thule_Society

If I run the simple following file with pdflatex, I get that \par output.

\documentclass[12pt]{article}

\begin{document}

\GenericWarning{FOOOOOOOOOOOOOOOOOOOOOO}

x

\end{document}

@PhelypeOleinik And also, no output from GenericWarning.

@CarLaTeX @egreg My husband says „Yes it is“. Then again it is „only“ one of many versions of the „sun wheel“ - quite a common symbol in celtic, norse and even indian culture. It got cribbed by the Nazis in an attempt to link their ideology to historical traditions. Whether or not it is to be considered evil very much depends on who uses it for what purpose.

@FaheemMitha Of course, all my statements above assume correct input :-)

@FaheemMitha \GenericWarning takes two arguments, so your code does the same as \GenericWarning{FOOOOOOOOOOOOOOOOOOOOOO}{\par}.

@PhelypeOleinik Ah, I was just wondering about that. Probably the same as \GenericError, so I've been using them both incorrectly for a while. Where is the documentation for those, please?

In the afternoon "A balcony at Siracusa":

@FaheemMitha The first argument is the "header" of the message. It's usually the package name printed between parentheses, and it's only printed when the message has more than one line. That's why FOOOOOO does not appear in your example. The second argument is the body of the message, in this case \par. The message ends with "on input line X."

@FaheemMitha Not sure about documentation. Give me a minute, I'll find it.

@UlrikeFischer @egreg Should we close the question?

@PhelypeOleinik I think I knew this once, then forgot. Someone here told me. I think I might have seen something that passed for documentation, too.

@FaheemMitha I could find it only in (texdoc) source2e p. 59... Not even clsguide has it... Perhaps it should (@DavidCarlisle,@JosephWright,@UlrikeFischer).

@PhelypeOleinik Ok. Well, thank you very much for the correction.

@FaheemMitha You're welcome :-)

@FaheemMitha That message should appear on the log too. You said it doesn't?

@PhelypeOleinik What message?

Is this the correct usage, then?

\documentclass[12pt]{article}

\begin{document}

\GenericWarning{foopackage}{FOOOOOOOOOOOOOOOOOOOOOO}
% \GenericError{foopackage}{FOOOOOOOOOOOOOOOOOOOOOO}

x

\end{document}

@FaheemMitha \par on input line 6.

@PhelypeOleinik That weird \par message does appear in both.

I was just complaining yesterday that TeX doesn't tell you the number of arguments.

@FaheemMitha For \GenericWarning, yes. Error messages take more arguments.

So that appears because I have the wrong number of arguments?

@FaheemMitha If you behave TeX behaves too :-)

@PhelypeOleinik Ok. What's the syntax for \GenericError?

@FaheemMitha Not the wrong number, but the wrong argument. The second argument grabbed in that case is \par.

@FaheemMitha I'm searching. I don't remember by heart

@PhelypeOleinik I intended there to be one argument. So TeX grabbed the first thing it saw as the second argument.

Is that a correct summary?

Sometimes I wonder everyone doesn't just go mad.

@FaheemMitha Yes

@FaheemMitha it has 4 arguments.

@UlrikeFischer Oh.

Can someone point me to the documentation, please?

This is something package writers use, right?

@FaheemMitha texdoc source2e.

@FaheemMitha no, normally package writers should use \PackageError.

@UlrikeFischer Oh.

@FaheemMitha well you have passed in \par as the message text, so what do you expect?

So, the GenericWarning info in that document says:

This takes two arguments: a continuation and a message, and sends the result to
the screen.

@DavidCarlisle Now, that's what I would have expected you to say. :-)

@FaheemMitha saying the truth can't hurt

So, "continuation"? What is that?

@CarLaTeX It certainly looks like one. Specifically the symbol used by some Panzer divisions in WWII, although reversed. But the image is of a branding tool, which would reverse it, so it seems likely it is.

@DavidCarlisle I feel I need a quote from Shakespeare at this point...

I wish I was better educated.

@PhelypeOleinik The doc says "continuation". So it's printed at the beginning of every line if it's a multiline message, and not shown otherwise?

Um. How do I get a multiline message, please?

@FaheemMitha that is exactly the issue we were discussing the other day, it is not that you passed the wrong number of arguments, \GenericWarning takes two arguments, it's just that the second argument (the actual message text) was \par

@DavidCarlisle Yes, I know. The point is that I intended to pass one argument.

@FaheemMitha \MessageBreak

\documentclass[a4paper]{article}

\begin{document}

\GenericWarning{zzzzz }{Some Message\MessageBreak Some Message}

blblb

\end{document}

@UlrikeFischer Ah, excellent. Thank you.

@FaheemMitha but latex can not check that, it can't know that you didn't want the message to be \par, it can not read English.

Ok, I see it at the top of that documentation page.

@DavidCarlisle Yes, I'm aware. Though some mechanism would still be nice.

@AlanMunn suggested something the other day, but I still don't understand the point.

@FaheemMitha I don't see how any mechanism is possible

@DavidCarlisle Well, I guess I will carry on seeing weird errors forever, and probably asking about them here. I'll try to be more careful to read the documentation about commands, but it's nice to have a safety net.

@FaheemMitha that was a suggestion that you could define a command that had arguments like foo(..)(..) so if there are not two () pairs you get an error, but for existing commands using standard tex {} arguments that isn't an option

@DavidCarlisle Is that useful for user defined commands?

Though regardless, I don't understand how it is supposed to work.

@FaheemMitha just define \def\foo(#1){\section{#1}} then compare \section with no argument with \foo with no argument.

@FaheemMitha no, as there is no nesting. \foo{{x}} will do what you expect, \foo((x)) not.

\documentclass[a4paper]{article}

\begin{document}
\def\foo#1{-#1-}
\foo{{x}}

\def\foo(#1){-#1-}
\foo((x))

\end{document}

@UlrikeFischer That's a great example. :)

@UlrikeFischer Although the "No" is perhaps a bit strong, depending on the particular use case.

@AlanMunn hm yes ;-). But I took the question as "generally useful for normal arguments" which it is simply not - you too often fell into the trap that the wrong end delimiter is used.

@UlrikeFischer Yes, I agree, as a general mechanism it's not really appropriate for sure.

Is there any advantage in using \GenericWarning, as opposed, to, say, just \typeout?

@FaheemMitha it is mostly the code that defines \PackageWarning and \ClassWarning, it is not usually called directly

@DavidCarlisle If I want something to return a warning, should I just use \typeout, then? I can't remember if TeX output can be directed to standard error, but it's not that important here.

@FaheemMitha either, if you use \PackageWarning it will be formatted like a latex warning, but if you don't mind about that \typeout is OK as well

@DavidCarlisle So I could use \PackageWarning or \GenericWarning?

@FaheemMitha package normally but there is really little difference, the definition of \PackageWarning is, in full:

\def\PackageWarning#1#2{%
   \GenericWarning{%
      (#1)\@spaces\@spaces\@spaces\@spaces
   }{%
      Package #1 Warning: #2%
   }%
}

@DavidCarlisle So one calls the other. Ok.

@DavidCarlisle I thought I'd probably asked that before.

@AlanMunn Indeed the tool is suspicious, I'll flag the question

If I'm calling \PackageWarning or \PackageError in the context of some macro, and it's not really a package, is it ok to use the macro name instead of the package name, or would that be too non-standard?

It's for my personal use, so I doubt it matters too much.

It would be nice if there was like a MacroWarning.

@FaheemMitha as I said last time you asked, if you have code that goes as far as making warnings then it ought to be in a package anyway but if it is in the preamble that is conceptually an "inline package" so \PackageWarning{faheem} or whatever you want.

@DavidCarlisle Actually, it's at the end of the file. I'm wrapping the call to a macro in a - if this file doesn't exist, then abort - style if then construction.

Possibly not the best design.

Maybe that check should be in the sty file. Let me see if I can move it there.

Yes, I think I'll do that. It's more sensible.