@cfr The linear order of adjectives in Welsh is basically the same as the linear order you find in English, but given that they are on the other side of the noun, you would expect a mirror image order, not the same order. That's the puzzle.

@AlanMunn Feel rather stupid but do adjectives have an order?

@cfr Yes, in most languages, for reasons we don't entirely understand, certain classes of adjectives tend to follow an order of combination: e.g. a big red American truck but not a American red big truck

Or do you mean in some cases?

@cfr As with everything language related it can get complicated quickly, and there are various ways you can temporarily adjust the canonical ordering for discourse related reasons, but the basic ordering facts are pretty robust, e.g. nationality adjectives typically combine before colour and colour adjectives combine before size (as in my example).

@AlanMunn So not all of them. I was trying to make it general, but then I thought of 'big blue' and wondered if you meant just a subset. It doesn't seem puzzling to me to put 'big' after the colour in Welsh. It sounds weird the other way ;).

@AlanMunn What I meant was there's not a complete ordering of adjectives. A partial one makes sense.

@cfr Really? At least the literature, a phrase like the truck example is supposedly "truck big red American". Is that not the order you would prefer?

@cfr And by prefer, I mean, roughly, if you were describing the object for the first time in a conversation, that's the order you would use. (To control for discourse messing up the basic order.)

@AlanMunn I'd definitely put 'American' last. I'm not sure about 'big' vs. 'red', but I'm not a first language speaker.

@cfr Ok. It's definitely tricky to have these judgements as a second language speaker, because these sorts of examples don't really arise that often in everyday speech.

@AlanMunn No, I think you're right. I'd just probably say it wrong. But if I wrote it, I think I'd realise.

@AlanMunn Plus very commonly combined adjectives in Welsh often have non-standard order: 'old cat white' for example, so there's no question of ordering the adjectives.

And it's easier with nationality words. They have endings which just feel as if they need to come later.

@cfr Ok. So the puzzle is, that the simple-minded analysis of "adjective on the right" would predict exactly the opposite order in Welsh: truck American red big, and since that's not the order you get, something else must be going on. So the idea is that in the same way that Welsh is VSO, where the verb appears initial in its clause (displaced from its object), it's also N initial in the noun phrase.

@AlanMunn What does N initial mean? Not just noun-first, obviously.

@cfr N initial with respect to the modifiers of the noun. Not completely N initial since obviously the definite article will precede the noun. So in the same way that a subordinating conjunction will precede the verb in an embedded clause, the determiner precedes the noun.

For example if you take a phrase like the driver's detailed description of the accident, the "subject" of that phrase (the driver) appears to the left of the noun, just like the subject in the clause The driver described the accident. In Welsh the order is description detailed the driver of the accident so the noun precedes the subject, just like the clausal order Described the driver the accident

@AlanMunn OK. But 'big red American' doesn't contain a noun.

@cfr It was truck big red American

@AlanMunn What I mean is: truck is the noun and that's first. But what does N initial have to do with the ordering among big red American.

@cfr So the idea is that the adjectives in Welsh actually combine on the left of the noun initially, which is why the linear order looks like English, and then the noun moves to the left of all the adjectives. This is the same way in which the verb appears displaced from e.g. its object in the Welsh clause.

@AlanMunn Ahh. OK. Thank you. Sorry for being slow.

@cfr No need to apologize. I do this stuff for a living. You just speak the language. :)

It doesn't work for modifiers, though, right?

@cfr What do you mean by modifiers?

Words which modify the force of an adjective e.g. quite or very. I'm sure there's a better term.

@cfr Ah, ok, yes, that makes sense though, because they are modifying the adjective not the noun. So it makes sense that they appear initial in the adjective phrase I think.

@AlanMunn But they don't. quite does but very doesn't.

cat white very vs. cat quite white. (I think.)

@cfr Oh really? That I didn't know. So you get tall very not very tall. That's interesting. I don't recall any data like that being discussed, but I haven't really been looking.

@AlanMunn Yes. iawn comes after the adjective. Definitely.

cath wen iawn - cat white very

@cfr What about other adverbs like unfortunately small or optimistically large?

@cfr It wouldn't totally surprise me if only 'very' was able to do this.

@AlanMunn No idea. Unfortunately - I'd say yn anffodus. But that's not exactly an adjective.

I'd have to look up optimistic ...

@AlanMunn Academy dictionary:

> Heb ganfod cofnodion cyfatebol wrth chwilio drwy'r prifeiriau Saesneg am gofnodion sy'n cyfateb yn union i'r chwilair 'optimistic'

@cfr But is there a way to use this adverbially to modify an adjective as in my examples with the suffix –ly added to the adjective?

@AlanMunn It does better with optimistically:

> yn obeithiol etc.

@cfr So not an adverb again?

@AlanMunn I ... don't know.

@cfr Fair enough. In my classes I teach "Alan's rule for adverbs: if you think it's an adverb, it's probably not" :)

@AlanMunn Exactly. gobeithiol is an adjective meaning hopeful. gobeithio is the 'berf enw'. I don't know what the English term is.

@cfr Google translate tells me this is "verb noun" which makes sense. Adjectives are exactly that sort of thing, sort of nouny and sort of verby.

@AlanMunn It's sort of an adverb, but not like an English adverb. It needs yn to do the job an adverb would do. Otherwise it's an adjective.

@cfr Yes, I understand. So it's more like English "with care" = "carefully" (of course 'care' is a noun but the idea is similar)

@AlanMunn Hmm... It translates as verb noun, but that isn't an adjective. It's the base form of the verb.

@cfr You've piqued my interest now. I'll do some digging in the next few weeks and tell you what I uncover.

@AlanMunn Yes.

@AlanMunn It's like to hope in English, but it isn't exactly equivalent. It's the form you find in the dictionary or you look up in your book of verbs or whatever.

I only mentioned it to explain where gobeithiol came from. I guess yn gobeithiol is parallel to yn gobeithio: the first gives you something adjective-like and the second something something-like.

@cfr Yes i got that.

@AlanMunn I'd be interested to know what a verb noun is in English.

@cfr But the placement of 'iawn' is interesting. So I'll see if anyone's written about that. Now though it's time for bed. :) As always, nice chatting with you.

@AlanMunn You're too patient! Good night.

@cfr That I can probably ask a couple of linguist friends who work on Celtic

TUG2024 is starting to look more healthy

@JosephWright Oh, very good!

@mickep no context talks yet - you should change that!

@UlrikeFischer I wish I could. But I'll go to BachoTeX and then to the ConTeXt meeting. I don't think I manage one more trip to the continent ...

@UlrikeFischer I was hoping Petr would attend ....

@mickep Fair enough

@JosephWright Yes, but a pity. I like Prague, and it was fun in Bonn...

@UlrikeFischer I could give it? I'm sure @mickep trusts me with all things context.

@DavidCarlisle Oh, I would not mind. You can run your examples on texlive.net.

@DavidCarlisle you could make a context version of xii.tex

The iso pdf describing the ua2 is not in pdf2 format but 1.7 or so, and if one dares to look inside the pdf file, one will notice that it is not beautiful... A bit like writing the luatex manual that can only compile with pdftex.

@UlrikeFischer I could, yes

@mickep We'd noticed :)

tinyurl.com/3ye43b84

@mickep ^^

@DavidCarlisle There is your talk

@JosephWright I find that highly disturbing.

@mickep To be fair, the idea of such standards to to define future tool features - so it's not surprising it's not itself a PDF 2.0 doc

@mickep well ISO is quite old fashioned (I don't have the final version so I can't check, but imho it is at least tagged). More problematic is that also the wtpdf document is not PDF 2.0, but at least they found a few members who could produce examples pdfa.org/wtpdf.

@JosephWright PDF 2.0 is seven years old. They could have managed ...

@mickep in case the irregular input was disturbing you tinyurl.com/bdwhh4ud

@samcarter maybe fontspec is loaded. But I'm not sure about that.

@Skillmon ... maybe they should load it a 3rd time, just to be sure :)

@DavidCarlisle Yes, now it looks more like the context documents I'm used to.

@JosephWright Looked like the ua2 document was created with InDesign. Adobe should be able to... (in fact the pdf includes opentype and type1 fonts in a not so optimal way, maybe Adobe don't care)

@mickep which one?

@UlrikeFischer ISO 14289-2:2024

@mickep ah, I don't have that.

$ grep -r 'north east lines' /usr/local/texlive/2024/texmf-dist/tex/generic/pgf
/usr/local/texlive/2024/texmf-dist/tex/generic/pgf/libraries/pgflibrarypatterns.code.tex:\pgfdeclarepatternformonly{north east lines}{\pgfqpoint{-1pt}{-1pt}}{\pgfqpoint{4pt}{4pt}}{\pgfqpoint{3pt}{3pt}}%

@cis ^^

@cis grep -r "north east lines"

@DavidCarlisle oh.

@Skillmon so slow

@DavidCarlisle back to dayjob (read in David Attenborough's voice: The defeated male leaves.)

@DavidCarlisle Ahhh.... pgflibrary....

@UlrikeFischer Essentially the same as the wu tang one.

@mickep I know ;-). That's why I didn't spent money on it.

@UlrikeFischer oh. Good choice.

@cis declare your own version

@daleif OK

@UlrikeFischer Something is definitely not happy. For example, with l3build check -cconfig-title title-006 in latex-lab things die with:

Package hyperref Info: Link coloring OFF on input line 12.
! You can't use `\relax' after \the.
<recently read> \c__pdf_object___tag/struct/1_int

:(

@JosephWright hm, I got that error too in certain combinations of the develop tagpdf. I will check with the ctan version.

@UlrikeFischer Thanks

@JosephWright ok I know what the problem is. the luatex backend of tagpdf assumes the older syntax to reference the objects in lua and now falls over. I do think that I can make a tagpdf update first (I added fallback code to switch between main and develop), but I need to run the tests first to be sure that this doesn't break.

@UlrikeFischer Cool thanks

@UlrikeFischer I'll build the l3kernel dev release, so it's ready-to-go once we know everything is sorted

@DavidCarlisle -- What? No Latin?

@barbarabeeton Contextus users non intelligunt classicas

@DavidCarlisle I'm surprised you did not choose the bible in latin.

@DavidCarlisle -- Quod responsio suspectum spectat ...

@barbarabeeton Ego oraculum infallibile consului

@mickep or hebrew

@DavidCarlisle -- Why not Greek? (Depending on which testament.)

@DavidCarlisle Indeed. I downloaded the Latin Nova Vulgata Bible, but not yet typeset.

@barbarabeeton Are you suggesting the ultimate authority in such things is not American? Shocking.

They also make for rather good language comparison examples.

@yo' -- you've got mail.

@barbarabeeton I saw, plan to respond later :)

@yo' -- Thanks.

@UlrikeFischer I don't understand the (UA2) tagging. How to mark the body of a description?

@mickep what sort of description? A list?

@UlrikeFischer No, for example "Theorem 1 bla bla bla bla. Here Theorem 1 ends"

The bla bla bla bla is the body.

It is difficult to impose a (nested) structure of things. Maybe I don't understand UA2, it seems to be meant for simple flat structure.

We put a Sect around the whole theorem, and a Caption around the Theorem 1. The rest is then normal paragraph or other substructures.

@UlrikeFischer Oh, but is Sect meant for that? The name suggests sections...

I'm very confused about this.

Putting Sekt around the theorem would surely make the theorem easier to understand :P

@samcarter Oh, indeed

@mickep well "meant for" is unclear, but in our opinion the most usable. We considered Part first, but the problem is that Caption is not allowed in Document and so also not Document>Part which means a theorem with Caption on the document level would be invalid.

@UlrikeFischer And a Sect in a Sect also gives problems, so ...

@mickep no, Sect can (and actually should) be nested

@UlrikeFischer Oh, really? Then I should try harder (got an error in the veraPDF validator)

@mickep what did it say?

@UlrikeFischer "<Sect> shall not contain content items"

@UlrikeFischer Do you nest Sect?

@mickep that doesn't relate to nesting, it means that you can't put directly text/MC-chunks in it, you need e.g. a P or some other structure first.

@UlrikeFischer Oh, that is not clear from the standard spec, but thanks!

Is that clearer in the WTPDF spec?

@mickep not really, the core parent-child rules are partly in PDF 2.0 (there is a matrix) and fully in ISO 32005. And no, I do not think that everything makes sense ;-)

@UlrikeFischer So, kind of a mess. The iso document I mentioned earlier today is more or less word by word the same as the WTPDF document. I want my money back! :)

@mickep you should have been at Dante, I said there that they are quite similar, apart from the price ... ;-)

@UlrikeFischer Different logos.

(What about copyright?)

@mickep that is ok, the document was developed together with the ISO group. See pdfa.org/resource/tagged-pdf-q-a

@UlrikeFischer ok, makes sense

@UlrikeFischer Still not clear on how to mark end of things, if possible. More trial and error...

@mickep what end of things?

@UlrikeFischer structures, like a definition or even levels of sections

On the bus now, can probably give example later

Am I right in my understanding that TeX reads \" (as an example) as a macro even if letters follow, because " has catcode other? So, is this why you don't need a space to tell TeX that the CS ends there?

@JasperHabicht Yes

@JasperHabicht Control symbol: escape char followed by exactly one non-letter

Control symbol, right! So, if I would tell TeX that b has catcode other, I would be able to use it in the same way, right?

@JasperHabicht in other news, in jsonparse you have a section 'Internal variables and commands', but these are documented code level interfaces - either you meant them to be __ or the section description is wrong

@JasperHabicht Correct

@JosephWright No, you are right, I should change the term there

@JasperHabicht I assumed so :)

I am happy to get feedback! I am not that deep in TeX terminology and I am also not sure whether jsonparse does things in an optimal way. I did test it but maybe I included some mean traps by accident due to my ignorance

@JosephWright Since JSON accepts \b and so on, I thought that it could be interesting to somehow parse these as well ... but I was unsure how to do that ... and I am still unsure whether this would be a good idea at all

@JasperHabicht BTW, please don't use 'author-maintained' - it leaves all sorts of questions if a maintainer ever goes AWOL (cf. biblatex, cquotes, ...)

@JasperHabicht Reading code now

@JosephWright So, just "maintained"?

@JasperHabicht Yes

@JosephWright Reading texdev.net/2012/05/04/the-lppl-maintainer-or-author-maintained

@JasperHabicht Personally, I'd avoid all the \JSONParseSet... commands and go keyval - you can then keep the storage internal

Yes! maintained it should be

@JasperHabicht \cs_new:Npn, etc., for expandable functions, so e.g. \jsonparse_parse_to_prop:Nn should be \cs_new_protected:Npn

@JosephWright Good idea

@JosephWright I have to admit that I am always unsure what to use when ... I need to refresh my memory

@JasperHabicht Anything that carries out an assignment, creates a group or does typesetting is not expandable

Simples

@JasperHabicht (In a csname, you could use \tl_if_head also - slightly faster)

@JasperHabicht Various document commands are marked as expandable but ... aren't

@JasperHabicht You've got a lot of e-type expansion - many look V-type or set_eq, others I'm not sure about - I think you want to check and review carefully

@JosephWright Yes, you are right. Now that you explained again where you use what, I also got that

@JosephWright The reason might be that I first started with a completely different approach. I'll check these again

@JasperHabicht Cool

<rant mode> Why on earth can I only paste images bigger than 160x160 px? </rant mode>

...on the main site, I mean.

@Rmano I don't know, it seems to have come in recently, someof mine it has rejected I was sure were bigger anyway

...pasting a lot of white pixels... 🤐

@JosephWright Thanks for your quick code-review!

What is the etiquette when an answer is accepted, but a somewhat better answer is provided literally years later?

@LaTeXereXeTaL switch the accept if you want (assuming you mean it's your question)

@LaTeXereXeTaL unless that means moving a tick from me to @egreg of course.

Won't that lower the original answerer's rep?

@LaTeXereXeTaL no lower than it would have been if the new answer had been posted at the time and you'd accepted that one.

@LaTeXereXeTaL the site machanisms are explicitly designed to allow newer answers to be added at any time, old posts don't "close" specifically to allow newer better answers to be posted.

@DavidCarlisle Okay. Both are great...ugh....

@JosephWright what has changed in latex-dev that could make compiling a dtx fail? I get an error in tagpdf-code:

without dev it works fine, but I want to compile the main documentation with lualatex-dev for the tagging ;-(

@UlrikeFischer v end of line handling changed

@UlrikeFischer and @DavidCarlisle, you got mail :)

@mickep clearly aimed at Ulrike since it requires actual knowledge not just bluffing

@DavidCarlisle Yes, maybe, but I added you since I got the feeling that you were the one who worked with that document. (There is [?,?] at some place, maybe a missing reference.)

@mickep well yes I sort of curated the collection but mostly by doing what I was told., I'll check the refs though

@mickep technical answer from @UlrikeFischer and a picture from me, take your pick:-)

@DavidCarlisle I see.

@DavidCarlisle Both were good, thanks!

@UlrikeFischer Thanks, I understood that was not all, but I did not see the "level structure". Will stare at it.

@DavidCarlisle grr. It compiles if I use the expl3-dev code, but then my tests will change ;-(

@mickep looks like the source has been fixed at arxiv, I just refetched a different bbl]#

@DavidCarlisle Ah, good!

@mickep fixed thanks

@UlrikeFischer I think what is still unclear to me is what can be nested in what. Maybe the 32005 document will tell when it arrives...

@UlrikeFischer Locally adjust l3doc

@DavidCarlisle Don't trust @DavidCarlisle! Obviously, if you've doubts, the green tick should go to me. ;-)

@JosephWright ah, it is enough to copy the new version?

@UlrikeFischer Yes

@JosephWright seems to work. So next try to build tagpdf ...

@UlrikeFischer :)

@Rmano They switched image hosting: meta.stackexchange.com/questions/395024/…

@JosephWright uploaded.

@samcarter i would understand a maximum upload size, but why a minimum? And so big? Strange.

@UlrikeFischer Thanks

@Rmano You could ask on main meta :)