twitter.com/TeXniker/status/…

@CarLaTeX yes, but.... picture mode or TikZ? :-P

@Rmano picture mode like @DavidCarlisle's airplane :D

Let's not talk about football :)

also quack

@PauloCereda ok, did you not hear that the FC Bayern München is sacking their trainer?

@Skillmon oh no

But I found that in München steht ein Hofbräuhaus :)

@PauloCereda did you also not hear that they now want to employ Tuchel?

@Skillmon ooh

@PauloCereda we rabbits are good at not talking about something :)

ooh rabbits are very skill[ed](mon) <3

@PauloCereda ask @egreg about it

@PauloCereda breakfast

@DavidCarlisle oh no

@DavidCarlisle ooh

@CarLaTeX ^^

@DavidCarlisle 😖

@DavidCarlisle 😂😂😂 not so aerodynamic as the benchmark one...

@Rmano is this better?

@DavidCarlisle well, barring the fact that you can remove three resistors and one capacitor from the circuit, it's not so bad ;-P

...and well, maybe a switch could be handy, but no one asked for a lamp that can be switched off...

@Rmano yes like the plane it's impressive it gets the syntax right

Meanwhile, texlive 2023 up and running ;-), finally

@Rmano ask for a switch and a switch is provided,er somewhere....

@DavidCarlisle 😂

github.blog/2023-03-23-we-updated-our-rsa-ssh-host-key

@DavidCarlisle someone should tell them about tools like docs.github.com/en/enterprise-cloud@latest/code-security/…

@samcarter got a bit of a shock when I got a security warning on push this morning, until I found that post :-)

@DavidCarlisle For a while you got the feeling that you were cut off from the team now that you only work in context... :P

@DavidCarlisle I'm a bit surprised that there was no mail or at least a banner on their website. For their move to 2FA, they sent mail, nag with a banner ...

@mickep who needs over-complicated formats like context and latex?

@DavidCarlisle Only we mortals...

@DavidCarlisle thanks!

@PauloCereda -- It's been there for a long time. I have my grandfather's beer stein from there; it's more than 100 years old, and the Hofbräuhaus is even older (founded in 1589). Sadly I don't like beer (except for Weissbier), but in asparagus season they serve very tasty lunches.

@barbarabeeton and ginger beer. 😋

@mickep -- Oh, yes! Especially when visiting Bermuda! (Stormy ...)

@barbarabeeton I bought a good one for tonight.

@mickep -- actually, ginger beer is my almost-every-evening libation. Not a coffee drinker (except for an occasional Turkish coffee or coffee with chicory). It's hot chocolate for breakfast. And tea at other times -- recommend "long jing" (also known as "dragon well").

@DavidCarlisle Great drawing as usual :D

@CarLaTeX it has other skills too:

@DavidCarlisle Ooohhh noooooo!!!!!!

There does not seem to be interest in standalone package interaction with algorithm package any more. here I am only using standalone to make single pdf to include as an image later on. I need to find another way to do this. By manual crop for example. Which not a big deal. Or by using geometry package.

@Nasser it doesn't really make sense to use floats with standalone, why include \begin{algorithm}[H] ???

@DavidCarlisle As I mentioned in the question, this was to fix problem with standalone. I put the link there also. But it fails using [H] or no [H] "However, if you give the algorithm the [H] option, it will no longer be floating, so you can use it." The above is from the answer linked to.

@Nasser I mean why have \begin{algorithm} at all? just as if you were doing table you would just have tabular not the surrounding table float

@DavidCarlisle I also tried that, and it gives error with standalone but not with article class. Please see

\documentclass[10pt,crop]{standalone}  %error
%\documentclass[10pt]{article}  %works
\usepackage{float}
\usepackage{algorithmicx}
\usepackage{algorithm} % ctan.org/pkg/algorithms
\usepackage{algpseudocode} % ctan.org/pkg/algorithmicx

\begin{document}

\begin{algorithmic}[1]
\Procedure{A}{}
    \ForAll{$a\in A$}
        \If{$a\geq b$}
            \State\Return $a$
        \EndIf

Same error mentioned in the question, "Something's wrong--perhaps a missing \item. \Procedure"

@Nasser well it's a vertical construct so you would need \documentclass[10pt,crop,varwidth]{standalone}

@samcarter just said same I see

@DavidCarlisle so slow :) Can you post an answer? Your explanation might be more helpful to future users than my "whenever there is a problem with standalone try adding varwidth"

@samcarter done:-)

@DavidCarlisle :thumbsup: :)

Could the babel and luatex folks look into tex.stackexchange.com/questions/680690 ? If the values after penalty= in \babelposthyphenation really affect the hyphenation in a way which is as counterintuitive for a user as shown, they are at least partially meaningless, either by specification (really bad!), or due to an error in the implementation (hopefully logically improvable), or due to a bunch of accumulated rounding errors (mitigation: a numerically better™ algorithm).

so apparently we are going to be struck by an asteroid this weekend and the world will come to an end. How will I get my context badge?

@AlMa0 they are not (ever) here, you could raise an issue on babel so it gets logged at least

@AlMa0 did you find more doc for babelposthyphenation than the sketch in the main babel manual? the argument is described as a Lua pattern, so I was trying to understand your |

@DavidCarlisle Frankly speaking, I do not understand the documentation at latex3.github.io/babel/guides/… . I am aware of the usages as in my issue report but not of anything else. E.g., I don't understand what no= is supposed to do. As for |, it is supposed to provide a hyphenation point. E.g., the hyphenation pattern “Zu-stands-ma-schi-ne” is written as Zu|stands|ma|schi|ne.

@AlMa0 yes or at least it seems to match the disc node generated by the default code, the babel manual could be clearer though:-)

@DavidCarlisle In my understanding, the first and the second argument of \babelposthyphenation are assumptions, and the third argument of \babelposthyphenation is an action that specifies what should be executed whenether LuaLaTeX encouters the second argument as a single word including all the hyphenation points (whichever they are, either built-in or supplied by either a package or the user via \babelhyphenation) in the language given by the first argument.

@AlMa0 I suspect what is happening is that babel's lua callback is adding disc nodes where you specify and these are then explicit disc that are always seen by the linebreaker, conversely not all the expected hyphenation points are triggered by luatex's built in pass (because luatex is luatex) so using babelposthyphenation makes more breakpoints available even if you only re-assert the same penalty

@DavidCarlisle Ohh, no

@mickep is that "oh no, we are all doomed" or "oh no, David may fail to get a badge" ?

@DavidCarlisle More the first to be honest. :)

@DavidCarlisle It seems to me that a disc (a “discretionary node”?) is more a implementation-level concept than a user-level concept. So if \babelposthyphenation is (by its specification or by the thoughts of the developer) supposed to perform actions on discretionary nodes, whichever these actions might be, then I as a user might be simply misusing a low-level command. (My fault in this case.)

Or is a “discretionary node”, on the contrary, is a user-level concept that a typesetter has to understand? In this case, I'm also at fault because I've not done proper reading.

@AlMa0 you are applying lua pattern matching to modify a node tree generated by the initial hyphenation pass, the documentation is a bit obscure but it is really low level modifications and not really anything easily explainable as a user command.

@DavidCarlisle Oh. I see, thanks!

@AlMa0 got your mail:-)

@DavidCarlisle I have not written anything to you directly. Or do you refer to github telling you that github.com/latex3/babel/issues/230 has been created/edited?

@AlMa0 yes that

@DavidCarlisle I guess, the issue report is mainly directed to the author of the interface of \babelposthyphenation (the roles of the interface developer and the programmer may but need not coincide).

@AlMa0 it doesn't quite work as I expected, but even if it had worked I think you would need to change your settings, changing a penalty by one or two is almost never going to change the output,

@AlMa0 sure, it's Javier, but we all get all the l3 issue notfications

@DavidCarlisle It does change change the output for me. However, the change is often unintended. So a change of the hyphenation of a word X might be just a side effect of \babelhyphenation{language of X}{X}{hardly matters what is written here}.

@AlMa0 if you have a long German word with 3 penalty-50 breakpoints and one penalty-49 one it does not mean it will usually prefer the 49

@AlMa0 yes you are seeing breakpoints that were skipped not be skipped, that is an observable change but not due to a numeric difference

@DavidCarlisle You probably mean \penalty49 would be preferred to \penalty50, don't you? (So, positive values, not negative ones.)

@AlMa0 well I mean they are effectively the same thing.

You want to move a break point from the current optimal place to somewhere a few characters away. Unless it is a very long paragraph this will usually involve increased stretch of white space. So you need to reduce the penalty for your preferred break enough to compensate for the extra stretched white space, and that is usually 10s not 1 or 2 penalty points

As for “almost never”: I have a list of 85 \babelposthyphenation entries in the preamble, one per word. In my document of 473 pages (without microtype) these entries incur 10 changes. (However, some words might have been removed from the document, but the \babelposthyphenation entries for these word are still there). The penalty= values in these entries vary between 47 and 52. Some paras are indeed long and involve math (which my also stretch and shrink), as in my issue report.

*which my also -> which may also

@AlMa0 as i say I suspect you are seeing the effect of that command adjusting the algorithm in unexpected ways more than you are seeing any effect from numerical values. That said there is always some example where a change of 1 makes a real difference, but....

@DavidCarlisle Got it!

@AlMa0 for a penalty change of 1 to make difference you have to be able to take that breakpoint with virtually no change to white space. it's not impossible but rather unlikely

@DavidCarlisle Alright. So let us view the issue report as a way to catch bugs in the specification of \babelhyphenation or its implementation. If the changes of penalty by 1 should not matter as you say, then the output should also not change at all or at least change in the intended direction.

@DavidCarlisle By the way, I think that as my penalty values vary between 47 and 52 in general, the effects of the \babelhyphenation entries on the words of the same paragraph sometimes interact, and I might get some of the 10 changes in the output due to more than one application of \babelhyphenation (whether it's for at least two instances of one word or for at least two different words).

@AlMa0 I don't suppose you would approve of this approach to the problem of long compound words? chat.stackexchange.com/transcript/message/19701043#19701043