@Skillmon yes (at least it makes the token memory available)

@Lupino the log shows the maximum usage, not if some was re-used

@Lupino no, I'd always advise against using it, personally. 999 times out of 1000 it's better to use XSLT and transform to tex, then you get a real xml parser and vastly more capable transformation possibilities.

@Skillmon that is why latex's \@onlypreamble exists (well that does \let\foo\@notprerr but it comes to the same thing)

@cfr Yay! <3 I got yours by the end of last week, thank you very much, it's adorable! <3 It's now being displayed on top of our Christmas tree! :) I poked you by e-mail a few days ago, but I am not not sure if reached you or if the spam filter blocked me (or even if the emails are correct). :)

@PauloCereda what kind of tree do you use for Christmas trees in Brasil? I suppose spruce trees are in short supply in those parts of the globe.

@HaraldHanche-Olsen we use artificial, plastic pine trees. :)

@HaraldHanche-Olsen Mostly everything is odd for us: pine trees, heavy clothing (we are in summer), snow. So we decided to play dumb and follow the trend of everyone else. :)

@PauloCereda so is there a heat stroke epidemic at Christmas, then? with all the heavy clothing, I mean.

@HaraldHanche-Olsen thankfully no, we just observe the rest of the world. :)

My family prepares the nativity scene, it's very nice. :)

@PauloCereda I don't bother with Christmas trees myself (having no children), but I have two tiny avocado trees in my office. They are amazing. I had tried for a long time to grow some, but in the end, these two came from some avocado seeds we threw onto the compost pile in the backyard.

@HaraldHanche-Olsen ooh guacamole trees. :)

(sorry couldn't resist)

awwwww

Lovely trees!

And there's snow outside!

@PauloCereda Indeed. No summer here. The sun is just coming up, and it's ten in the morning.

@HaraldHanche-Olsen brrrr

@HaraldHanche-Olsen nice! How much up north yo are?

@HaraldHanche-Olsen nice! How much up north you are?

@yo' 63°25'

@HaraldHanche-Olsen wow! That's reall up to the north. I'd like to visit some place around that, I just can't decide whether to go in winter or summer.

@yo' Summers can be nice here. It never really gets dark, although we're not quite in midnight sun territory. The best part of winter is aurora borealis, but those are unpredictable, plus you have to go yet further north for a better chance at seeing it.

It would be nice if texlua had a -e option. Currently it doesn't. At least in my version.

faheem@orwell:/tmp$ texlua -e "print(_VERSION)"
Script file print(_VERSION) not found
faheem@orwell:/tmp$ lua -e "print(_VERSION)"
Lua 5.2

Would it be reasonable to post this somewhere as a feature request?

@HaraldHanche-Olsen well, I would possibly go to Oulu, which is yet a bit up (65 00) and a.b. is one of the motivations

(and Finnish sauna every day is the other one :) )

@yo' Including a refreshing bath in a nearby lake, I presume? That's not quite for me (in winter). Lack of sisu, I suppose.

@DavidCarlisle quack

@PauloCereda brunch!

@DavidCarlisle oh no

Hi folks. The Debian Lua packages are in a horribly unmaintained state, but in many cases, uploading new/fixed versions would not be hard. In fact, quite easy. Does anyone here know of any Debian Developers who would be willing to sponsor Lua package uploads?

@HaraldHanche-Olsen yep of course! Without outdoor refreshing water, sauna is not a real sauna!

@PauloCereda wouod cerainly enjoy that :) ^^

@yo' ooh :)

@DavidCarlisle unfortunately, xslt is way too slow for efficient print production.

@DavidCarlisle if the overall sum of used memory is the same with and without \let\<csname>\undefined, then no memory is "freed", or is it? How would one test it?

@Lupino have you tried compiled version eg saxon EE? but in any case xmltex's highly limited translation abilities are related to that speed as it is parsing line by line and outputting things as it goes so apart from the limited grab feature it is only suitable for very simple structures where the source structure matches the final output very closely

@Lupino classic tex doesn't use dynamic allocation so at the system level the memory usage is essentially constant, all allocation is handles internally by tex itself, if you redefine a macro eg by \let it to another command or to undefined then the memory that was used to store the replacement text becomes available for future definitions. That is why latex undefines many commands at begin{document} to leave space for \label and \ref in the document.

@DavidCarlisle my recent questions about a "global version of \csname…\endcsname" from a couple of months ago were the result of my attempts to bypass those limitations. And i made some promising progress parsing and rendering html tables with css classes and style attribures, since. Moreover, i think by now that xmltex's functionalities could be extended vastly to cope with more complex xml structures.

@DavidCarlisle, @barbarabeeton, @egreg What's the deal with \phi vs \varphi and Unicode? The MATHEMATICAL ... versions all have the symbols reversed: I'm confused!

@JosephWright you really should read my documents:-)

@JosephWright w3c.github.io/xml-entities/#phi

@DavidCarlisle I'm working on the Unicode 'de-math' idea, and getting the Greek data together is painful (also deciding how to handle \mathnormal: I guess like Will I'll allow for TeX/ISO/French/upright/....)

@DavidCarlisle Thanks: I'll preuse later

@DavidCarlisle I think I can do expandable math-mode-to-Unicode and give something half-useful

@HaraldHanche-Olsen Re this question of mine – I tried it, and it worked fine at least to do about 3000 tex.print(…) inside a single \directlua{…} without hitting any sort of memory constraint. (Each line being on the order of 30 characters.)

@yo' Just don't forget to cut a hole in the ice first. And don't linger in the sauna too long, or the hole will freeze over.

@DavidCarlisle There's a whole question in my mind about how much 'symbol' stuff can go into (in principle) generic expl3 ... really can I use \alpha or whatever? I think yes, but ...

@HaraldHanche-Olsen you can always dive into snow :)

@JosephWright please call the iso one iso80000 or some such, it's so annoying to have it described as iso as if it's a universal standard for math layout when it's some style guidelines for some small section of engineering as far as I can tell.

@DavidCarlisle Er, let me work out Greek letters first ;)

@DavidCarlisle I'll need to modify my plan a bit: version 1 just has \mathnormal hard-coded

@DavidCarlisle I'm wondering about all the shape weirdness, whether to go with what Will does in unicode-math, or just go 'tough, the shapes should map Unicode; this is fallback data', or ...

@DavidCarlisle Tidying everything up is ... hard work

@JosephWright The “closed phi” is mostly used in the US. Elsewhere, the common choice is the “open phi”. Not sure what's common in Britain.

@egreg Oh, great; not even a consistent choice

@egreg Don't mathematicians have something like IUPAC to decide on these things?

@JosephWright I have been shuffling data in unicode.xml since 1999 :-)

@JosephWright Are you joking? ;=)

@egreg No: I'm thinking of something like iupac.org/projects/project-details/?project_nr=110-2-81

@DavidCarlisle :)

@JosephWright one possibility to get more math structure into a plain text rendition of mathematics is this unicode.org/notes/tn28/UTN28-PlainTextMath-v3.1.pdf which currently is more or less just MS Office but might get a higher profile...

@JosephWright Every mathematician is jealous about their own notation. There are some “standards”, but of course they differ by nation.

@DavidCarlisle It certainly will if it's what I implement for hyperref-2 ;)

@DavidCarlisle Looks more-or-less what I was thinking anyway ...

@egreg Hmm, that's extremely frustrating. Ah well

@egreg I'll probably just go with how unicode-math handles the \var.. shapes: it's then at least consistent between TeX and the 'text output' I'm aiming for

@JosephWright it's the "linear format" in word so implemented so you can actually test the back translation, which has some advantages over ad hoc linearisations.

@DavidCarlisle Yes, true

@DavidCarlisle I'll get as close as I can: it's a Unicode doc, so there is the whole 'this is not just us going off on one' business

@DavidCarlisle Also gives me a keyword list I can stick to

@DavidCarlisle I guess the keywords are all the same as unicode-math, but I will check ...

@DavidCarlisle I wonder how Chris will react to this :)

@DavidCarlisle I still have to work out how to extract operator names ...

@DavidCarlisle I guess I might have to do asciimath too ...

@AlanMunn -- I managed to drop my macbook when packing up to come home from the Palo Alto TUG meeting. But it was in its padded case, and it dropped on my toe, so it was the toe that took the damage. Not broken, fortunately, but seriously purple for weeks, and now finally growing a new toenail. I hope you didn't lose any important electrons.

@AlanMunn oh no :(

@HaraldHanche-Olsen -- I'm glad that you didn't leave those two little trees on the compost heap -- they would have frozen by now. Yarn bows in pretty colors make very nice holiday decorations. They don't even have to be tied around the branches. Just tying bows separately and balancing them on the leaves gives a very cheerful effect.

@HaraldHanche-Olsen -- And I bet the sun goes down about two in the afternoon.

@barbarabeeton That's a good guess. It's just after four now, and it's dark outside. And has been for a while.

@yo' -- I'm not sure he'd really be fond of chopping through the ice.

Just a quick query here before I ask a question on the main site: In luatex, if I say token.set_macro("foo","\\\\") then the expansion of \foo shows as BADBAD. Why is that? (I tried giving it an explicit catcodetable argument, but it did not help.)

Oddly, searching the main site for [luatex] set_macro did not produce a single hit.

@HaraldHanche-Olsen -- It's easy to remember that noon divides "ante meridiem" from "post meridiem". (Practicing my Latin. Last night we attended the annual Latin Carol Celebration, and sang lots of seasonal songs in Latin, including some that originated in that language. But not "Duodecim Dies Natales"; think "xii".)

@barbarabeeton So @DavidCarlisle should have named that file duodecim.tex …

Hey everybody, greetings from Göteborg in Sweden. Here it is dark since a while as well...

(")>/

@barbarabeeton Back around 1981 when I visited Toronto for a year, there was a conference in Waterloo. A bunch of us people from Toronto went back one day for a party. Getting onto the freeway, we were suddenly confused abut whether we were going the right way. It was about noon, I looked out and saw the sun on the right, and said “it's all right, we're heading east”. A car full of mathematicians did not believe me! I was shocked.

@JosephWright -- Unicode starts with the premise that the primary form is the one used in the language text. Knuth started with the assumption that the base form is the one used most commonly in math. Unicode didn't exist until long after TeX was in use, and there weren't any mathematicians on the relevant standards committees (those were mostly engineers, with a few important linguists).

@egreg -- Very true. We should just be grateful that the UTC was willing to accept the principle that math is a genuine, open-ended, language. It was a struggle.

@HaraldHanche-Olsen -- Well, "xii" is a roman numeral. This discussion is about on the order of that about \phi vs. \varphi.

@barbarabeeton ;)

@barbarabeeton @PauloCereda well, the question is, can ducks swim in a 4C/38F water?

@HaraldHanche-Olsen setting a catcodetable works for me (but I'm not sure what you want as content):

\documentclass{article}
\usepackage{luatexbase}
\directlua{
token.set_macro(luatexbase.registernumber("CatcodeTableOther"),"foo","\\\\")}
\show\foo
\begin{document}
abc \foo cde
\end{document}

@yo' brrrr

@HaraldHanche-Olsen -- Well, the current day length is just under nine and a quarter hours, and since we're near the eastern edge of the time zone, it starts getting dark a bit after 4 p.m. Actually, sunset has been at the same time for about a week, and will start getting later (very slowly) on the 19th.

@UlrikeFischer Whoops, I had just asked on the main site. I'll check out your solution (or is it a workaround?). Not really familiar with luatexbase. I had tried a catcodetable created by \initcatcodetable1, which should be the catcodes of initex. That did not help.

@barbarabeeton Ah yes, the good ol' equation of time up to its usual tricks.

@yo' -- I think eider ducks can cope with water at that temperature. (I simply stop breathing when I get into water much colder than 72F/22C. And I'm not really comfortable if it's less than 83F/28C. So the sauna routine doesn't really tempt me.)

@PaulPaulsen hard to say in general you don't want to use T1 fontenc as that forces 8bit legacy tex-specific font encoding and one of the main points of luatex is to use opentype unicode fonts. but it may be that your docuemnt is using packages setting up fonts that are only available (or only available under the names used) in tex encodings so you may need changes elsewhere,

When is the UK election? This Friday?

@PauloCereda thursday

@DavidCarlisle oh twelve.

@PauloCereda lords-a-leaping

@PauloCereda Huh? FridaY??? What the heck??

@barbarabeeton Sure, but the glyphs marked 'MATHEMATICAL' could presumably be 'the right way around'

@DavidCarlisle and a kipper in a EU-regulated package hanging in a tree. :)

@JosephWright oh

@barbarabeeton They have heat exchangers in their legs. Pretty nifty if you ask me.

@DavidCarlisle Okay, thank you! I will see if I find out who's responsible for this ;)

@HaraldHanche-Olsen -- I didn't know tht. Yes, very nifty!

@barbarabeeton I wish I had them. My toes get cold easily.

@DavidCarlisle ooh

@JosephWright -- Before proposing that (not saying that's a bad idea), check to see what Microsoft Office is using; almost certainly the same as what's in Unicode TR#28, which I haven't checked, but which I think you said is "backwards" from the TeX point of view.

@DavidCarlisle -- Yup. Watching swans landing on ice is even funnier.

@barbarabeeton Word follows TeX

@barbarabeeton Handily have the latest Word on my laptop: desktop at work doesn't

@JosephWright boo

@PaulPaulsen if you show the error in the log, probably easy to debug

@JosephWright -- Take a look at unicode.org/reports/tr25 page 9. It has some exposition regarding the history of phi in Unicode.

@UlrikeFischer Sort of works, but you need the lua code in a separate file to avoid tokenization of the \\ . I tried putting it in a luacode* environment instead, but that did not work: It left \foo undefined.

@HaraldHanche-Olsen luacode is an environment, so you would have to make foo global.

@UlrikeFischer sounds of hands slapping forehead

@UlrikeFischer Actually, I think that luacode really ought to avoid that. It has to capture its body anyhow in order to pass it to \directlua, and it should be feasible to do some group trickery to place this code outside the environment. Though by now it probably can't be done for backward compatibility.

@HaraldHanche-Olsen personally I avoid luacode altogether I can never remember which characters it changes the catcode of and which versions are grouped and using \directlua almost always seems simpler

I get that, but luacode* is promised to change every catcode to other. But it seems not to be robust enough. I find it easier to just put all lua code in an external file, just calling it with extremely simple \directlua invocations.

@HaraldHanche-Olsen yes a separate lua file solves a lot of problems:-)

@DavidCarlisle I suppose I could also use \begin{filecontents*}{foo.lua} … That's robust enough for the purpose, isn't it?

@HaraldHanche-Olsen yes probably (writing it out to a file and reading it back isn't the fastest thing though)

Whoops, I wasn't thinking … I had a palindrome within reach, but asking a question ruined it. Or was it accepting the answer that did it? I should learn not to do that.

@HaraldHanche-Olsen you could unaccept mine for a bit

@DavidCarlisle :) Actually, I have been ten points from a palindrome for ages. Usually, points trickle in often enough that I should have had it by now, but lately, nothing. I have been avoiding answering questions because I could easily blow past my upcoming palindrome. (And also because I've been too busy.) If another ten points come in, I may well try your suggestion. Meanwhile, the green tick stands.

MY DUCKS HAVE ARRIVED YAY

@PauloCereda ? which ducks?

@UlrikeFischer he probably ordered a Chinese takeaway dinner

@UlrikeFischer the duck nativity one :)

@DavidCarlisle I don't think there's chinese takeaway in Analândia.

@DavidCarlisle oh no

@yo' :)

@PauloCereda I'm crap with email, but I'll look. We've been on strike, so there are backlogs ....

@cfr oh no worries. :)

@bp2017 I don't think there any exact definition. What is usually referred to as the font size should really be called the design size. Typically, it corresponds roughly to the height of a capital letter or the width of the letter ‘m’. But that is all up to the font designer.

@bp2017 Oh hang on, the height of a capital letter plus the depth of a typical descender is closer.

@bp2017 Even that comes up a bit short: For the standard font in tex, “10pt” computer modern, a typeset box containing the text “Tj” is 6.83331pt high and 1.94444pt deep according to tex.

@barbarabeeton My MacBook Pro has been dropped numerous times with no ill effects other than some dents on the case. They're pretty bullet proof. But the iPad is much more delicate, because the screen is not surrounded by anything. I'm so glad I got AppleCare+ on it which allows for two instances of accidental damage in addition to the regular hardware coverage.

@bp2017 it is a number made up by the font designer

@yo' yelp.com/search?cflt=chinese&find_loc=Anal%C3%A2ndia+-+SP

@AlanMunn -- Also, an iPad is much lighter than a Macbook, and wouldn't do nearly as much damage to one's toe There are some reasonably attractive cases, and we'v got those. Has been really useful for Gordon, who keeps dropping his in the driveway. But they have both gotten pretty tattered after a few years. I guess I just didn't grow up as part of the "disposable" era.

@barbarabeeton I'm actually pretty careful, and this is the first major damage I've caused to any of my computers over many years.

@AlanMunn -- Being careful is always a good policy. But it isn't infallible, unfortunately. Do continue to take care.

@barbarabeeton Yeah: my XPS has a dent in the corner :(

@PauloCereda At work, we've got money for a workstation ... looking at the MacPro ...

@PauloCereda ... but the budget is not that big (we are thinking £2k+)

@PhelypeOleinik I just wanted to say thanks for the advertisement of my message, wasn't necessary, but thanks anyway :)

@Sigur amsart does its uppercasing with \uppercase without expanding the argument, and since \uppercase doesn't change control sequences, the parts of the title which are stored in a macro aren't upper cased.

Enrico's next desktop:

apple.com/uk/mac-pro

@JosephWright that strange cheese grinder?

@Skillmon :)

@Skillmon Oopps

@JosephWright ? Is something wrong with my explanation?

@Skillmon Nope: the oops is about amsart

@JosephWright but they take care to not uppercase maths.

@Skillmon Quite doable with @DavidCarlisle's textcase package, or today the expl3 case changer (being upgraded at the moment, and possibly replacing \MakeUppercase Real Soon Now)

@JosephWright if allowed to be non-expandable, not upper casing maths isn't that hard, two simple loops and you're done (one to ignore $...$ and one to ignore \(...\)) -- as seen in amsart.cls :)

@JosephWright care to summarize it?

@Skillmon Sure: it's a bit more tricky to do expandably, and to cover all of Unicode ;)

@JosephWright what's Unicode? :)

@Skillmon Well @DavidCarlisle's code is pretty short; mine does an expandable loop over everything, checking each token. The current code uses one loop, the plan is to move to a two-loop plan (one to expand, one to case change)

@Skillmon github.com/latex3/latex3/blob/l3text/l3kernel/l3text.dtx

@Skillmon :)

@JosephWright my favoured couple of lines from that:

% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}

@Skillmon :)

@Skillmon It's quite straight-forward, really :)

@Skillmon Check each token, see if it's an implcit char, see if it's a macro, otherwise see if we can case change it

@JosephWright and provide some ad-hoc equivalents of font changing macros, yes, I saw that, still it's just much.

@JosephWright may be I should do an in depth comparison in in the style of this

@DavidCarlisle :)

@DavidCarlisle funny answer, but it should be a comment, imho.

@Skillmon I made it a comment first but actually, if trying to persuade a co-worker which package to use, a statement from the author of one of them is probably a more useful answer than a technical assesment

@DavidCarlisle as a comment this would still be valid, and as an answer it is of limited use for future visitors. But that's just my opinion and a bit of reasoning how my opinion was formed :)

@Skillmon imho it is quite useful. Many people simple want to know what they should do and not why.

@JosephWright as-- While this is true, a public update to amsart shouldn't be expected soon, unless management policy changes. There will probably be an illusory update that merely provides a mechanism for specifying the 2020 version of the math subject classification.

@UlrikeFischer are there that many Windows users on this network :)

@HaraldHanche-Olsen, @DavidCarlisle, thank you.

@JosephWright @UlrikeFischer I just pushed a ltxcmds branch of hyperref with two commits, one avoiding ltxcmds and one etexcmds, I think it may be worth doing (in which case it probably needs some review) or maybe we should wait until expl3, but I think this may help gradually reduce the hyperref dependencies

@Skillmon, @JosephWright, thanks for discusison and explanations on upper case.

What should I do to fix that?

I really would like to insert a macro in title argument.

@Sigur Can you guarantee that it is expandable? If so, you could do \makeatletter\xdef\@title{<your fully expandable title>}\makeatother instead of \title{<your fully expandable title>} (and do you need a short title, too? In that case you can use \xdef\shorttitle{<your fully expandable shorttitle>})

@Sigur you could use textcase package and then \let\uppercasenonmath\MakeTextUppercase

@DavidCarlisle I installed it and will try it.

@UlrikeFischer thanks

@Skillmon, thanks. To be true, I only would like to insert a personal name in title, using a macro instead of typing it directly. The name is being used many times in text via a macro.

@DavidCarlisle it stills load ltxcmds, is this expected?

@DavidCarlisle, thanks. Let me see what I can do.

@UlrikeFischer oops I was testing something and put it back for a while, give me a minute...

@DavidCarlisle the date is still at 2019/11/10 v7.00c right?

@UlrikeFischer ah I have %\RequirePackage{ltxcmds}[2010/11/12] but it's included by pdftexcmds

@UlrikeFischer yes I didn't bother with dates or real doc yet

@UlrikeFischer next on my list of things to write out was the tex part of pdftexcmds (I need the lua part) same as done in epstopdf, I guess I need that sooner rather than later.

@UlrikeFischer @JosephWright that makes it a bit more work, so question is, is it worth doing, or should we use expl3....

@DavidCarlisle I mistook the closing brace of (HO) for the one closing of pdftexcmds.

@DavidCarlisle I would simply add \RequirePackage{expl3} and go along ...

@UlrikeFischer trouble is that then you have to more or less re-write everything rather than just find things like \ltx@ifundefined and change it to \@ifundefined, so you can more or less mechanically avoid the dependencies.

@DavidCarlisle, your solution worked except if I use \\ inside title (I know it is not a good pratice but I'd like the value of \foo in second line)

@UlrikeFischer my thinking (I think, it keeps changing) is that in teh end we'll want to use expl3 for low level conversions and xomputations but for etexcmds which is just checking if you can use \unexpanded we should simply write it out and use \unexpanded and for most of the ltxcmds ones we would use 2e versions lile \@ifundefined as it's basically a 2e package

@Sigur hmm that should probably work in textcase, how come it's not robust which version of latex have you?

@Sigur this works for me?

\documentclass{article}

\usepackage{textcase}

\begin{document}

\MakeTextUppercase{aaa\\bbb}

\end{document}

@DavidCarlisle, the same for me.

> Microsoft Teams is the First Office App For Linux

@Sigur ah this:

\documentclass{article}

\usepackage{textcase}

\begin{document}

\centering
\MakeTextUppercase{aaa\\bbb}

\end{document}

@Sigur that works in latex-dev but not latex as the centering version of \\ is only robust in latex next, so you need \protect\\ or use pdflatex-dev

@DavidCarlisle Yay!

@JosephWright you see the important latex-dev feature is fixing @egreg's \\ bug, your little extra feature of including expl3 goes unnoticed:-)

@DavidCarlisle ooh

@DavidCarlisle I'm missing something. If you don't load etexcmds who is defining \etex@unexpanded?

@UlrikeFischer same issue, I think I'm too tired let me get rid of etex@ as that's easy, but then I think I'll get rid of pdftexcmds and infwarerr (did that already on epstopdf) and then see how it looks. probably not finish that this evening.

@DavidCarlisle ah, kvsetkeys.

@UlrikeFischer that one is tricker as I'd like to use keyval or l3keys but both are likely to have user-observable differences in edge cases. I suppose I could pull kvsetkeys out of oberdiek and stop it loading etexcmds but not using it would be better, makes me oscillate back towards the big expl3 bang approach....

@DavidCarlisle Yup

@DavidCarlisle and we want to remove pdfdoc encoding ...

@UlrikeFischer Sounds good to me

@UlrikeFischer I should have Unicode math sorted, er, by next week

@UlrikeFischer yep but I wanted to separate structural re-arrangements from actual functionality changes.

@DavidCarlisle but how much can still be rearranged without affecting functionality?

@UlrikeFischer all of pdftexcmds/ltxcmds/etexcmds/infwarerr can certainly be written out just to use normal 2e coding, then as I say for the rest (encoding changes and pdf form stuff etc) we should use expl3

@JosephWright @UlrikeFischer I'm wondering if we should have a hyperref-core that avoids contrib packages, uses expl3 and could move towards being in a future format and have hyperref load that but have all the "patch third party package" stuff.

@DavidCarlisle next secret plan ;-)

@DavidCarlisle Sounds good to me

Can one write \\ with \csname ... \endcsname? (Luigi suggested that as a workaround for the token.set_macro problem).

@UlrikeFischer \csname\@backslashchar\endcsname

@DavidCarlisle oh yeah, 55 miles!