NAA spotted while browsing ^.^ I don't intend to make an account so have at it

@Skillmon @barbarabeeton ooh twitter.com/RabbitEveryHour/status/1525692755527204866

@PauloCereda @UlrikeFischer Presumably the correct response is "Good" and close with no action? github.com/latex3/hyperref/issues/235

@DavidCarlisle ;-). I'm just checking, but I wanted to handle Karl's issue first.

@DavidCarlisle oh no

@DavidCarlisle can I blame @PhelypeOleinik? :)

@DavidCarlisle Why's that happening?

@JosephWright it's most definitely not my fault

@UlrikeFischer by the way you did the last update as ho-tex rather than l3-team?

@PauloCereda be quick, as it'll soon be lunchtime

@DavidCarlisle I do like \ShowHook :)

@DavidCarlisle isn't hyperref still ho-tex? (I sometimes confuse the mail addresses, but ctan didn't complain)

@JosephWright I like it too ;-)

@UlrikeFischer I think we are putting down as latex3 nowadays (it's realistic)

@UlrikeFischer :)

@JosephWright I will try with the next update and ask ctan.

@UlrikeFischer we moved the git repo from ho-tex to latex3 organisation

@UlrikeFischer oh no

@DavidCarlisle oh no

@UlrikeFischer @samcarter ^^ ooh

das Zebra

@PauloCereda :D

@DavidCarlisle @JosephWright well the problem is that hyperref loads nameref in the middle of the begindocument hook. I'm no sure that I really want to find out, what exactly is happening here, imho it is bad to load a package there. So I would prefer to require nameref normally.

\documentclass{article}

\AddToHook{begindocument}[something]{}
\DeclareHookRule {begindocument}{something} {after} {hyperref}
\AddToHook{begindocument}[hyperref]{\RequirePackage{nameref}}

\begin{document}
Foo
\end{document}

@PhelypeOleinik @DavidCarlisle @JosephWright the hyperref issue boils down to the fact, that hyperref loads nameref in begindocument, and nameref contains a hookrule too. So this rules reproduce the issue:

\documentclass{article}

\AddToHook{begindocument}[something]{}
\DeclareHookRule {begindocument}{something} {after} {hyperref}
\AddToHook{begindocument}[hyperref]{\DeclareHookRule{begindocument}{showkeys}{before}{nameref}}

\begin{document}
Foo
\end{document}

@UlrikeFischer looks like delaying is to check if typexml or tex4ht are loaded in the preamble, but I'm sure tex4ht can cope these days, and no on has used tyepxml since 1997

@UlrikeFischer hyperref has 11 places it tests for typexml :(

@DavidCarlisle is it still not my fault, right?

@DavidCarlisle yes, but it doesn't make much sense, why should/can nameref be loaded at the end of the package if typexml is used but only in begin document otherwise? (I will check if tex4ht has a problem if nameref is loaded, but if yes it would make more sense to ask tex4ht to adapt. )

@UlrikeFischer No it doesn' load it if typexml is used, and it delays the test until begin document:

    \ifHy@typexml
    \else
      \Hy@CatcodeWrapper{%
        \RequirePackage{nameref}[2012/07/28]%

@UlrikeFischer but I don't think typexml is available anywhere (I can't find a copy and I wrote it)

\ifHy@typexml
  \Hy@AtEndOfPackage{%
    \RequirePackage{color}%
    \RequirePackage{nameref}[2012/07/28]%
  }%
\fi

@DavidCarlisle ^^^

@UlrikeFischer blame @PauloCereda

@DavidCarlisle ?

@DavidCarlisle oh no

@DavidCarlisle do you think I can remove all these tests?

@UlrikeFischer I am pretty sure typexml here does refer to an email I sent Sebastian one day with a sketch that later became xmltex. Unless you know of anything else of that name,

@UlrikeFischer why not?

@DavidCarlisle I'm jalous

@UlrikeFischer /hug

@UlrikeFischer after lunch blame will need to revert to you

@DavidCarlisle oh no

@DavidCarlisle github.com/jayphelps/git-blame-someone-else

@DavidCarlisle and could it be that documents using xmltex set the typexml option?

@UlrikeFischer I just searched xmltex and jadetex sources and dont see "typexml"

@UlrikeFischer oh you mean they set it explicitly, perhaps but I'd still remove it from hyperref, anyone still using xmltex will be able to patch something in.

@DavidCarlisle ok. I found the old jadetex discussion, and there hyperref was called with \RequirePackage[implicit=true,colorlinks,linkcolor=black,bookmarks=true]{hyperref} so I have some hopes that Norbert will not complain ;-) So I will remove this tests and code.

@UlrikeFischer Probably because the upload email doesn't seem to be used for ID verification at all. All the upload form says about it is "This email address will not be published. It is meant for contacting you in case of questions."

@DavidPurton hi mr koala!

Hi @PauloCereda!

I've been absent in the tex.sx world for a while! Been fixing NeoMutt and can't do more than one hobby at a time

Hmmm. How can I tell what is happening with hyphenation for Australian English with babel?

@gusbrs According to ctan.org/file/help/ctan/CTAN-upload-addendum it is used for such verification.

@DavidPurton How do you mean "what is happening"? Are you looking for the hyphenation patterns? See the results?

If I make Australian the default it doesn't seem to hyphenate very well at all

If I make british or american the default and load australian as a second language it follows whichever one of the other two is main

I'd expect australian to use british hyphenation

but this seems not to be the case

@DavidPurton show an example.

@MarcelKrüger True, I had missed that.

oh - it doesn't change with which other language is main - it just doesn't hyphenate with austrlian

So maybe my question is really: How do I make australian use british hyphenation and is this a bug?

Ha! Maybe I just need to pass hyphenrules=australian british to \babelprovide

Ah - now I see the differences in babel-en-**.in. GB has ukenglish, US has usenglishmax and AU has english.

I think you must use the import key, without it babel tries to guess the hyphenation from the name austrialian: \babelprovide[import=en-AU,main,language=Default]{austrialian}

@DavidPurton ^^

@UlrikeFischer That does help (don't know how I missed that), but still doesn't match british.

I think I'll run with \babelprovide[import=en-AU,main,language=Default,hyphenrules=ukenglish]

@barbarabeeton bjoernkarmann.dk/occlusion-grotesque

@PauloCereda I hope they asked the tree for permission.

@mickep Maybe it's the MIT (might inquire a tree) license

@PauloCereda ... so instead of the weight, one specifies the age of the font?

@samcarter pretty much this :)

@PauloCereda :)

@PauloCereda oh no

@DavidCarlisle I found a \ifHy@typexml in passivetex (in fotex.sty).

@UlrikeFischer grr I checked sources for xmltex and jadetex but I guess I missed passivetex. Remove from hyperref and if necessary I'll claim write access to passivetex (I almost certainly wrote that bit anyway:-)

@DavidCarlisle I have now reinserted the \newif in hyperref. So it will be always false but at least not error directly (there is also a package from Don Story which uses the conditional).

@UlrikeFischer You could remove the tests but leave the option to set the boolean

@DavidCarlisle I think it would be good if it errors if someone really uses it. Better than silently doing nothing ...

@UlrikeFischer looking a passivetx i wouldn' be nothig, hepackag is saying "if hyperref bookmarks are being used wih xmltex" so the user could still use the option, wouldn't change hyperref but passivetex would take that branch

@DavidCarlisle hm yes. But not in hyperref. I removed the stuff there, and if that is needed for passivetex too, it could be half broken couldn't it?

@UlrikeFischer I don't think it is (but didn't check) may be able to check later (if I can remember how to use fotex) but if you leave tthe option in, fotex will always be able to test for it and pach in whatever it needed before)

why do people accept answers at exactly the wrong time?

@DavidCarlisle ooh

@DavidPurton oh, is NeoMutt worth a switch? Using good old mutt here.

@Skillmon ooh rabbit mail <3

@PauloCereda thought it's a dog?!

@PauloCereda oh, a carrot!

<3

@Skillmon woof :)

@PauloCereda ooh, ducks are good at foreign languages!

@Skillmon ooh

@JosephWright you have been saved

bbc.co.uk/news/uk-england-derbyshire-61437488

@DavidCarlisle :)

@DavidCarlisle wait what

@DavidCarlisle We mortals do not usually complain if our answers get accepted. :P

@samcarter ooh le canards

DAVID TENNANT WILL BE BACK IN 2023

@MarcelKrüger I came across tex.stackexchange.com/q/373866/52406, and tried it in MetaPost (no ode solver, so just following the flow). It is fast! 2-3 seconds with the same number of lines.

@mickep How do you mean by following the flow?

@MarcelKrüger Just doing Euler steps.

So no Runge-Kutta, but that would be pretty simple to do, I guess.

(I haven't looked what put-ode does)

@mickep I'm not experienced enough with numerical methods to really place that I fear (My mathematical background I more around set theory) but I'm not very surprised that it's much faster for this. We get a lot of overhead from emulating PostScript which MetaPost should be able to avoid since there "only" the rendering has to happen in Lua. I guess that implementung the ODE solver in Lua should be even faster.

@mickep According to the doc: "Package pst-ode uses the Runge-Kutta-Fehlberg (RKF45) method with automatic step size control for integrating the differential equations."

@MarcelKrüger I see, thanks. So a slightly smarter RK.

@MarcelKrüger Runge-Kutta-Fehlberg sounds like a cheese name. :)

@mickep Anyway I think it would be great if you could add a MetaPost based answer to that question. IMO MetaPost is severely underrepresented on the site.

@PauloCereda Oh, cheese

@MarcelKrüger I could probably do that, yes. I feel a bit bad adding answers using the non-requested tool.

@PauloCereda better name here (although not quite same thing) nag.com/numeric/nl/nagdoc_latest/flhtml/d02/d02pef.html

@MarcelKrüger I arrived to the question from tex.stackexchange.com/q/644238/52406. That one seems harder. I mean, we can draw many of the solution curves apparently, but it is not so easy to get them placed that well (as Mathematica does).

@DavidCarlisle ooh numerical cheese

@DavidCarlisle Oh, fortran :)

@mickep the language of choice, naturally

@DavidCarlisle Naturally. :)

@mickep it's a syntatic error to program fortran without a blue tie :)

@mickep I was hoping to get a Lua interface to that so you could call from luatex, wouldn't have to be so different to nag.com/numeric/py/nagdoc_latest

@PauloCereda Oh no, I need to go back 25 years to our computer labs, and correct...

@mickep :)

@DavidCarlisle Now you are nagging, aren't you? :P

@mickep It is an embedded method, which means it is basically 4th order, with one extra calculation to get a 5th order error estimator, allowing to adjust the internal step size according to the change in the solution. This allows one to get get the same precision for whatever number of output points chosen.

@AlexG So, more clever than just doing Euler steps.

@mickep So to say.

@AlexG Since I will probably not implement rungekutta45 in lua (probably done somewhere), I wonder what one gets in pstricks by just doing taking a time length dt and steping with dx = f(x,y)dt, dy=g(x,y)dt.

@mickep 1st order (Euler) vs 4th order (RK) precision. Visually the same, probably, provided small enough dt.

@AlexG I found pjb.com.au/comp/lua/RungeKutta.html. Remains to see if I can convince Hans to add support for it...

ooh more cheese

@gusbrs twitter.com/kangfrancoliu/status/1526250200863543298

@PauloCereda Oh, that would be an improvement for some people I know! :)

@gusbrs I knew it. :)

I've written my dissertation in LibreOffice, and my previous experience with the MS counterpart left me terribly afraid of file corruption. So I would do a local copy every hour... Desperate times. :)

@UlrikeFischer having fun at ctan?

@DavidCarlisle All the fun

@DavidCarlisle ;-). But I disabled the option for now. If this is really a problem we can readd it, but then I would like to have an example so that I can also check if some of the hyperref code was relevant.

@UlrikeFischer yes if it breaks passivetex I guess I will have to adopt it and adjust things there. Really makes no sense having that many typexml code branches in the main hyperref code.

@DavidCarlisle I'm happy about every branch I can kill ;-). The tex4ht branches are quite down already.

@UlrikeFischer :)

@JosephWright every time I look at it I find something I can remove, e.g. text mentioning version before 2003 or so ;-)

@UlrikeFischer :)

@UlrikeFischer There's a lot of history there

@UlrikeFischer Speaking of which, I'm going to ask about \expanded again soon ...

@JosephWright you have my vote

@UlrikeFischer 2003 doesn't seem so long ago. ask Matthew

@DavidCarlisle ooh future maintainer

@PauloCereda passivetex-for-paulo-to-fix-all-the-issues-in-his-copious-spare-time

@DavidCarlisle :)

@DavidCarlisle I suggest to add a new branch, passiveaggressivetex.

@mickep er well.. there is some history to that name.....

@DavidCarlisle Oh

@mickep there is a hint to the origin of the name in the last verse of the poem at tug.org/TUGboat/tb20-3/tb64ogawa.pdf

@UlrikeFischer might sympathise with the last line of the first verse:-)

@DavidCarlisle Oh, hm, I don't get it, but thanks anyways.

@DavidCarlisle I should hang it on my wall

@MarcelKrüger ^^^ our new table water

@UlrikeFischer Yeay, I always wanted to be a saint.

Saint Marcel, the saint one prays to in case of font problems :)

@samcarter In Portuguese, a water spring is called a "fonte", which is the same word used for font.

@gusbrs Then it is very fitting that this saint seems to sell water bottles in his merchandise store :)

@Skillmon my impression is that NeoMutt is more actively developed. Mutt seems mainly to be getting bug fixes. The NeoMutt main developer is very responsive and also very willing to accept contributions. I improved its composing of multipart/alternative emails and previewing of multipart/related emails. Text email is great, but emails need to look good on phones now. Even my own sent mail needs to be readable on a phone, so need to be able to easily send HTML alternative.