If @PauloCereda lived in Michigan, he could have a duck on his licence plate.

@AlanMunn ooh

That's nice!

@AlanMunn oopsie. :)

Fixed. :)

@AlanMunn: AAOOA is almost IOWA :)

@PauloCereda Hmm. Maybe with a southern accent. :)

@AlanMunn :)

@Alan: will you attend TUG 2019?

@PauloCereda Where is it?

@AlanMunn Palo Alto, CA

@PauloCereda Not too likely, but I do have old friends in San Jose. They were just asking if I was coming out to N. CA any time soon.

@AlanMunn Cool. :) I actually considered going, but you know the prices and the visa thingy... :)

@PauloCereda Yes, indeed.

@PauloCereda I guess @AlanMunn will be better in me in explaining how this then lead to the term "accountant". a modern version of a vampire.

@marmot lol That's a great folk etymology! When I was in grad school we used to play that as a dinner party game: who could think up the most convincing fake etymologies of words. The things that linguists do for fun...

@AlanMunn Wait, are you saying there is another reason why they are called accountants? ;-)

@marmot My PhD advisor (who is very, very wealthy) once said "Semanticists are like accountants. Everyone needs one, but who would want to be one." Needless to say we poor grad students didn't quite relate to the "everyone needs one" part. :)

@AlanMunn I see. Are you a semanticist? ;-)

@marmot Nope. Although I work at the interface between the syntax and semantics (as did my advisor). And much of the experimental work I do with children is really about how children learn semantics.

@AlanMunn You are chatting with an adult who probably never learned it properly. ;-)

@marmot Of English semantics, maybe. Of German, I'm sure you managed just fine. Not to mention Marmot in which you are clearly fluent.

@AlanMunn Yes, marmot language is much more structured. (No I seriously never thought about these things after having been "blessed" with 7 years of Latin at school.)

@marmot Sure, but conscious knowledge is overrated. It's the unconscious knowledge that's the interesting stuff.

@AlanMunn I guess it is nontrivial to determine what the unconscious knowledge is. How is it defined?

@marmot Well it's the implicit rules that govern the regular behaviour of language. So just like any other hard to see things you build a theory of what it would be like and then show using indirect methods that linguistic behaviour is reasonably modeled by that theory.

So for example, I can build a fairly simple theory of German word order which accounts for the fact that main clause word order is much more flexible than embedded (subordinate) clause word order. We can verify this to a large degree just by asking you which sentences you can say and which you can't. Or asking 30 such speakers in an experiment (although for lots of judgements, you don't really need many speakers since there's no appreciable variability.)

@AlanMunn Yes, but how do you determine it is a knowledge when it is subconscious? What I mean to say is how do you say that someone acquired a knowledge of some rules as opposed to, say, remembers some phrases? Or is knowledge in that context more broadly defined?

@marmot Well you have an infinite ability to produce structures you've never heard before, so it's implausible that you have a list. So you need minimally abstractions over properties of the phrases you hear.

@AlanMunn OK, so if you are able to reproduce a pattern then you would say this is knowledge? (I am really just wondering because the term "knowledge" seems to be differently defined defined over the disciplines. I agree of course that the physicist's definition is a bit at odds with statements like "Person A knows how to ride a bike".)

@marmot Well for a start, would you call your mental list of morphemes (minimal units of meaning paired with some pronunciation) knowledge?

@AlanMunn What are "minimal units of meaning"? (Sorry, I am really that dumb. ;-)

@marmot You know for example, that the word 'marmots' has two parts, one that means whatever 'marmot' means, and one that means (roughly) 'more than one'. So 'marmots' contains two morphemes, 'marmot' and '-s'. So does 'unhappy'. (un-happy). But you know that in the word 'under', 'un' is not anything.

@AlanMunn Yes, but this is a collection of rough rules and their exceptions. What is minimal about this?

@marmot ??? I wouldn't call these rules (or exceptions). The string pronounced 'marmot' in English has no smaller parts with respect to its meaning, whereas the the word 'marmotologist' has 3 minimal parts with respect to both sound and meaning. That's what I mean by minimal.

@AlanMunn No I mean that s indicates plural and un negation are rules, and there are exceptions.

@marmot Put another way, 'marmotologist' arguably contains the words 'marmot' and 'marmotology', but the word 'marmot' contains no other sub parts, and so is minimal.

@AlanMunn OK, now I got it, I think.

Minimal information unit, I mean.

@marmot I'm not sure what you mean by exceptions: there are words for which the pronunciation of the plural is irregular (like 'people' or 'sheep') but we would still want to say that 'people' contains two units, not one.

@marmot Sure, that's fine as a paraphrase.

@AlanMunn Yes, I meant the example of child and the like, where you do not put an s.

@marmot Right, but we still need to say that 'children' has two minimal units w.r.t meaning, even though the sound part doesn't line up very well. German is especially evil in this respect.... :)

@AlanMunn Is Mandarin also treatable with this systematics? (Yes, what I wanted to say is that there is a rule "To indicate plural append an s, but there are exceptions to this rule.)

@marmot Of course. But not every language decides to turn the same bits of the cognitive conceptual space into linguistic units. So in Mandarin, we don't have a morpheme that indicates plural (at least not in the way that English or German does.) But the concept of plurality still exists, it's just that Mandarin doesn't express it directly with language. But it has numerical quantifiers like 'many' or 'most' and obviously it has numbers.

@AlanMunn I see. Yes, language and related things are tricky. Many people would say that it is difficult to understand special relativity but without thinking they would immediately be able to tell you that, say, K is before R in the alphabet, which requires IMHO much more complex operations.

@marmot Some of these sorts of things are quite subtle crosslinguistically. For example in English, the sentence "All the doors will not open" can mean "Not all doors will open (some will) as well as "none of the doors will not open". But the German equivalent only means that none of the doors will open.

@marmot Yes, exactly. Cognition is super complicated, but we take most of it for granted because it's so automatic and unconscious.

@AlanMunn Actually not really, or depends how you translate. If you say "Es werden nicht alle Türen aufgehen" then the meaning is like in English. The more tricky example is "Du musst Dein Zimmer nicht aufräumen" which, if translated literally, give "You must not clean up your room" which is very different from the German meaning "You do not need to clean up your room.".

@AlanMunn Yup, and there are some autistic people for whom some GR computations are as simple as figuring out that a K comes before an R.

@marmot Yes I was thinking of a more literal translation, something like "Alle Türen werden nicht öffnen."

@AlanMunn Even this is ambiguous. In the south several will interpret it like the English version.

@marmot Really? That's interesting. I thought it was unambiguous in German.

@AlanMunn Not at all. There are huge differences between the regions and misunderstandings rather common.

@UlrikeFischer This is a bug. I will look into the gs sources and try to solve the problem. But it may take a while on the Artifex end to accept a patch.

@AlexG I'm cursed. Everytime I look at some package the first time I find a bug ;-). Btw: As I saw dvips (and I think xdvipdfmx too) always creates a named object for the second argument of bdc, so probably like pdfbase the driver code should do it for pdftex/luatex too to get a consistent result (and this probably means that we need to sort out how the resources are managed ;-(). dvips uses "R+number" for the name. Do you know why they decided for this naming sheme? @JosephWright

@UlrikeFischer "R" as in "Resource" ;-)

@UlrikeFischer with ps2pdf, resources are managed automatically. They are autom. added to the `/Resources << /Properties << /R1 10 0 R /R2 11 0 R ...>> >> dictionary of the corresponding content stream (Page object or XObject). With the other drivers, this must be done manually. When doing manually, one must detect the kind of content stream one is currently is currently inside (Page or XObject). I did this in pdfbase.sty.

@AlexG yes I saw this. Did you ever test how this resources are handled when the pdf is included in another one?

BTW the naming scheme is completely arbitrary. For the other drivers I used /rm@oc<int>.

@UlrikeFischer This shouldn't be a problem, as resources are local to the content stream. They are mapped in the /Resources<</Properties <<...>> >> dictionary of the stream object. Any embedded PDF is converted to an XObject (a self-contained stream object like the page object) with its own local resources.

@AlexG well the question is what happens if two packages uses the same naming sheme. I mean if tagpdf or expl3 would use /rm@oc<int> too we could run in problems ;-)

@UlrikeFischer That's why I'm trying to get this into order

@AlexG yes, I think too.

@UlrikeFischer driver.<name> for expl3

@JosephWright we will need a number too.

@UlrikeFischer i'll work on it

@UlrikeFischer I probably need to adjust a bit: I went for l3... for stuff, but then all of the PostScript code needed more order, so most of that is now driver.<thing>

@JosephWright that sounds a bit like some secret language ;-). The dvips/dvipdfmx code for the emc/bdc stuff is easy, as it automagically adds the objects. Here is it really pdftex/luatex where one has to look into.

@UlrikeFischer I'll look into it

@UlrikeFischer With (x)dvipdfmx the dict that is associated with a BDC tag must be manually added to the stream resources, as with pdftex/luatex. Only ps2pdf does it automagically.

E. g. \special{pdf:put~@resources~<</Properties~<</rm@oc1 10 0 R>>>>}.

But indeed, the correct resources dict @resources to which to add is detected automatically.

(Here we don't need to find out ourselves, whether it is the page resources or the XObject resources dictionary.)

@AlexG has dvips a special too to add something to such a resource manually?

@UlrikeFischer No. Only by brute force: e.g. [{ThisPage} <</Resources<</Properties <<...>> >> /PUT pdfmark which would overwrite an existing Resources dictionary. Not a good idea!

@AlexG well if one would keep track of the current content of the dictionary and rewrite it it could work (I'm thinking about how to get more or less similar working commands in pdftex and dvips. If we have a command that appends something to the /Properties object in pdftex, it should do something sensible in dvips too -- not quite sure if it is really possible in all cases in the end.)

@UlrikeFischer How will you keep track of the names (/R<int>) and associated object numbers that are automatically generated by the BDC pdfmark? That is not possible.

@AlexG oops.

@AlexG then probably the relevant commands will have to be noops in dvips.

@UlrikeFischer Any marks you want to add to the content stream must be added via the BDC pdfmark. This should work for all possible needs. Yes, a no-op would be an option for dvips

BTW For ps2pdf, there is no \pdfliteral{...} counterpart available as pdfmark.

(In order to insert arbitrary stuff to the content stream.)

@AlexG which means that there is currently no way to add BMC ;-) (not so tragic, it is only used in artifacts, and one can use BDC with a name instead)

@UlrikeFischer As for BMC, I will file a bug report and provide the patch on the GS bug tracker. If we are lucky they accept it shortly.

@AlexG good. Does dvips/ps2pdf/gs have an option to create uncompressed pdf? I couldn't find something yesterday and had to use pdftk to inspect the pdf.

@UlrikeFischer ps2pdf -dCompressStreams=false

@DavidCarlisle Or use ps2pdf12, as it makes v1.2 PDFs

@UlrikeFischer I am using pdftk too, as a VIM plugin. Thank you @David and @Joseph!

@JosephWright yes but then you might loose some newfangled funky pdfmark object thingies (not sure when they were all added)

@DavidCarlisle Most were added around 1970 or so ;-)

@DavidCarlisle doesn't work for me.

@AlexG isn't some of the tagging stuff newer?

@UlrikeFischer worked for me (there is another one to not compress font)

@DavidCarlisle I get this ^^^

@UlrikeFischer 易於閱讀

@DavidCarlisle ;-)

@DavidCarlisle The relevant BMC, BDC, EMC, MP, DP pdfmarks were defined already in the Pdfmark manual for Acrobat 5 in 2001.

@AlexG ah OK so if I last read the pdf/mark manual in 1998 I haven't missed too much:-)

@DavidCarlisle Perhaps even in the manuals for earlier Acrobat versions. I just don't have them on my book shelf.

@PauloCereda So the only thing I do well is "take genuine breaks"

@DavidCarlisle bingo, ps2pdf -dCompressStreams=false -dCompressPages=false test-utf8.ps worked

@PauloCereda and "eat well"

@AlexG the "coffee mug mat" on my desk is an acrobat 3.01 CD, I could see if it is still readable after a few years of ill-treatment, I suppose it has a manual on there as well as the software:-)

@DavidCarlisle Really??? That's kind of relic you must treat with the respect it deserves, from now on!

@JosephWright Do you have some time to help me l3build PGF? Here is what I have so far: termbin.com/eofg

@HenriMenke I'll take a look

@JosephWright Thank you!

@HenriMenke Have some 'real' work to do ... may take a little while

@JosephWright You're already doing me a big favour. Take your time.

Dear Paulo Cereda,

Congratulations, your upgraded Web of Science ResearcherID is now available on Publons, part of the Web of Science Group.

Hmmm.... OK?

@CarLaTeX That's not fair, Italians do always eat well. :)

@PauloCereda Yes, I'd seen that too

@JosephWright Is that any good to us?

@PauloCereda No idea

@PauloCereda Any idea what kind of ducks those are?

@HenriMenke ooh

@HenriMenke I'll probably need to extend l3build to support pgf: to-date, we've only had packages that install into one part of the TeX tree

@PauloCereda They are very colorful. I spotted them in the river at the university. I've never seen them there before though.

@HenriMenke Mallards, perhaps?

@HenriMenke They are very beautiful!

@JosephWright Don't overburden yourself. PGF is probably a bit esoteric, especially because it also supports plain TeX and ConTeXt. I mainly wanted to use l3build for regression testing.

@PauloCereda Exactly! The only things I do well according to your scheme!

@HenriMenke Right yes: regression testing alone perhaps a little easier (but even then ...). We've managed to support a range of set ups, it's just a question of adding a few more features

@PauloCereda I found them. They are acutally native New Zealand ducks. en.wikipedia.org/wiki/Paradise_shelduck

@HenriMenke ooh! :)

@JosephWright I definitely need a testsuite for PGF. Everytime I touch something, everything else breaks right now.

@PauloCereda @PauloCereda is that as useful as researchgate and academia.edu ?

@HenriMenke Got it!

@DavidCarlisle I... I've no idea!

@JosephWright That was quick.

@HenriMenke No, I mean I get the concept ... getting the code working going to take a bit longer! (Currently I'm marking essays)

@JosephWright Essays? I though you're teaching chemistry or have you switched to English literature?

@HenriMenke Yes, essays: first year work on a range of topics. We have to train people to write reports.

@JosephWright ooh trains

@PauloCereda the two main local news: The newspaper has a competition to find three hidden rubber ducks, and the cyclist club is driving to amsterdam to get lots of chocolate.

@UlrikeFischer OH MY

@PauloCereda they are claiming that they are doing it to prove that one can transport goods by bicycle, I think they are doing it to get lots of chocolote ;-)

@UlrikeFischer :)

@DavidCarlisle ^^

@DavidCarlisle ^^ GAME OF THE YEAR

There's this mozzarella guy from GitHub who keeps spamming inboxes.

I have a question.

Why the Wikipedia on mozzarella tells that the texture is semi-soft? Couldn't it be semi-hard? :)

This is quite unnerving.

@UlrikeFischer Bug report and patch submitted: bugs.ghostscript.com/show_bug.cgi?id=700984

Now let's wait and sip tea.

@AlexG Cool

@AlexG Thanks

@AlexG, @UlrikeFischer OK, so I'll start on a generic driver interface for marks

@AlexG BTW, at some stage I hope the generic driver code will cover all of pdfbase (though I'm still thinking about how far to go in e.g. supporting older PDF viewer bugs)

@JosephWright That was easy.

@AlexG Yeah, I got that, but it's a question of how long such fixes remain 'around'

@JosephWright Thanks! I am looking forward to it.

@AlexG May take a little while ...

@AlexG Oh, BTW, which Adobe Reader versions have that bug? The link in your source doesn't seem to say when/if it was fixed

@JosephWright You mean, how long it will take to accept a patch? Don't know. The responsible knows my name, fortunately. Let's see.

@AlexG I think Joseph means comments like "%dumb dummy Widget, workaround for AR bug;" in your sty.

@JosephWright The AR bug has never been fixed. So the current version continues to be affected. I gave up reporting AR bugs. They never reply to it. But, the workaround works well, hopefully forever. I found it by chance.

@AlexG I'd better add that then!

@AlexG Oh right

@UlrikeFischer Now this is the stuff we need to get into the kernel: we really do have to have a single structure

@UlrikeFischer This is what I'm imagining we can fix at a physical meeting: you, Frank, me to sketch something

@JosephWright yes, but which one? We have here a dictionary with subdictionaries.

@UlrikeFischer @Joseph This dummy widget annot ensures that the main, interesting annotation that was placed on a PDF layer by adding the /OC ... entry acknowledges the visibility setting of the layer. This doesn't work in other viewers at all. See: tex.stackexchange.com/a/477884

@UlrikeFischer This is what we need to sketch out! I guess we need a 'create a dictionary' function plus 'add to dictionary'

@UlrikeFischer This is only needed for implementing hyperrefs ocgcolorlink option for links. In ocgx2, I implemented a replacement, that also wraps around line and page breaks. See: tex.stackexchange.com/a/361422

@JosephWright my enthusiastic contrafibularities!

/wink wink

@JosephWright ok, we will have to think about it. Looking at Alex code and at tagpdf: we both obviously need an absolute page counter: \int_new:N\g_pbs_page_int %abs. page counter, \int_new:N \g__uftag_abspage_int.

@AlexG Reminds me I better add the 'no nested link' stuff to the dvips driver code

@AlexG I know, but it was meant as an example that not only one package wants to write into the /Properties dict.

@UlrikeFischer Ah, ok1

@UlrikeFischer Like I said, we want something like \driver_pageprop_dict_new:n (a bit wordy with pdf in it!)

@UlrikeFischer Perhaps \driver_pdf_pageprop_new:n (implicitly a dict) then \driver_pdf_pageprop_gset:nnn <dict> <key> <value>? We'll need some 2e hook, but that can be \cs_new_eq:NN \@pdf@pageprop@new \driver_pdf_pageprop_new:n

@AlexG, @UlrikeFischer I better start a list of jobs: AR fix, dvips nesting, BDC/EMC, PDF catalog, ...

@AlexG the discussion started in part because of this: tex.stackexchange.com/a/253417/2388

@JosephWright Frank would draw a mindmap ;-)

@UlrikeFischer I think I know what we want, it's just a question of exactly where to put it

@UlrikeFischer 'Load expl3 in the kernel ...'

@JosephWright we could make a branch expl which installs in latex-expl ;-)

@UlrikeFischer I'll need to do something like that

@JosephWright probably there should be no user command to create a new dict, but \driver_pdf_pageprop_gset:nnn should check internally if the dict is already in use.

@UlrikeFischer Ah, there is that

@UlrikeFischer Do we know in advance all possible dictionaries?

How to make fonts in \begin{equation} environment a little bigger for a special equation

@JosephWright In pdftex and luatex one must watch out for the correct Properties dictionary. If we are currently writing to the content stream of an XObject instead of the current page (by placing BDC into a TeX savebox that is then passed to \pdfxform to make the XObject) then \pdfpageresources is NOT the right command to employ.

@AlexG AH, right: more complexity

@JosephWright Unfortunately! dvipdfmx is smarter here as @resources named object is always the right thing to be written to.

@AlexG Oh, right, I see

@AlexG Because it's working from a complete document?

@JosephWright Don't know. dvipdfmx somehow keeps track of the current content stream that is open and being written to.

@JosephWright the possible entries in the /Resources dict seems to be a closed list (ExtGState, ColorSpace, Pattern, Shading, XObject, Font, Procset (deprecated in 2.0), and Properties). And happily all (apart from the deprecated Procset) have dictionaries as values.

@UlrikeFischer Ah, right, in that case yes we just need 'add to dict'/'remove from dict' (?)/'clear dict', plus probably a 'all pages'/'current page' split (plus issue @AlexG mentions)

@JosephWright That is, if you execute \special{pdf:put @resources <</Properties <<...>> >>} inside an open TeX savebox which is later distilled to an XObject by \special{pdf:bxobj ...}\box\mySaveBox\special{pdf:exobj ...}, the properties thus added end up in the /Properties <<...>> dict of the XObject, not in the one of the page.

In a step toward its goal of building out a data science development stack for web browsers, Mozilla today detailed Pyodide, an experimental Python project that's designed to perform computation without the need for a remote kernel (i.e., a program that runs and inspects code). As staff data engineer Mike Droettboom explained in a blog post, it's a standard Python interpreter that runs entirely in the browser. And while Pyodide isn't exactly novel -- projects like Transcrypt, Brython, Skulpt, and PyPyJs are among several efforts to bring Python to browsers -- it doesn't require a rewrite of…

Ew (better)

@JosephWright I guess we need "clear" for the /Resources dict (for the next page) and create/append for the subdictionaries.

@PauloCereda I avoid python since the one day I had to debug a python script which had messed up the spaces and the tabs ... I want real braces.

@UlrikeFischer :)

@UlrikeFischer here we go again. :)

@UlrikeFischer LOL

@JosephWright Those resources that belong to the page are collected in a L3 seq for each page and written at Shipout (writing to aux file is also necessary). For the other ones a stack is mainatined that keeps track of the open XObjects (see xsavebox.sty, command \xsb_addto_props:n) . Oh my, it is complicated...

Yes, XObjects can be nested ....

@AlexG why do you need to write to the aux?

@UlrikeFischer Everything! To explain things I did a while ago ... I try to prepare an example document.

@UlrikeFischer Because things need be collected to be written at shipout.

@JosephWright going for a walk. :)

@AlexG yes but why the two-pass through the aux? I need to write tons of labels to it (with pdftex), because I have to renumber the markers by page, but for the resources I would have assumed that using a \write would be enough.

@UlrikeFischer Page related resources must be collected, then written at shipout, finally wiped when the new page begins. Exact page breaks are only known after a first LaTeX run. And we don't want to have OCG check boxes on the viewer's "Layers" navigation panel, when there is actually no OCG present on the current page being displayed.

Such checkboxes persist, if we don't clean page resources.

We interrupt this TeX-related discussion to bring THE GEESE PARADE:

@UlrikeFischer ooh tihs is better :)

@AlexG you already got an answer to the bug report ;-)

@JosephWright @UlrikeFischer Resolved and fixed: git.ghostscript.com/…

@Student404Mus you haven't said what the question is but using any size command inside math mode is an error, \large and friends should be before the equation not inside it.

I've just find out that pepperoni pizza is "pizza al salame piccante" and not "pizza ai peperoni" (= pepper pizza). What a funny translation!

@CarLaTeX as we know, you still have a lot to learn about pizza:-)

@DavidCarlisle Pizza al salame piccante is good, no pineapple on it :P

@CarLaTeX a fast search for "pizza salami ananas" suggests otherwise.

@UlrikeFischer oh no!

Thank you @DavidCarlisle for your care and service. I'd like to mention that I wrote in the same manner and it works fine

@Student404Mus check your log vvvv

aTeX Font Warning: Command \huge invalid in math mode on input line 7.


LaTeX Font Warning: Command \LARGE invalid in math mode on input line 7.


LaTeX Font Warning: Command \Large invalid in math mode on input line 7.


LaTeX Font Warning: Command \large invalid in math mode on input line 7.

@Student404Mus the warnings are there because they do not work.

@PauloCereda don't spoiler things.

@Student404Mus we don't add warnings for no reason:-) if you use a font size command in math mode then it may or may not work depending on lots of details (in particular there you are using array so every cell is actually a new $...$ math list internally so the \large is not in the same math list as the math. Why put it inside the equation (where it doesn't work) rather than before it?

@Student404Mus also that is an equation not an array of values so you should use aligned not array so it is set in display math.

@Skillmon Sorry mr. rabbit

@JosephWright sorry

@PauloCereda do you like the latest commit here github.com/u-fischer/putfenc?

@DavidCarlisle You are 400 points away from a nice palindrome. Don't get downvoted!

@marmot been a long while since some vim users downvoted me before and pushed me off a 4-9 cycle on to a 5-0 one, but means I can aim for all the 5's after having got 444444 and 333333

@DavidCarlisle Let's keep our <s>fingers</s> claws crossed.

@UlrikeFischer -- Good stuff, that! I assume that Bär gave his approval;.

@barbarabeeton we didn't drink it, we got the foto from a friend ;-)

@UlrikeFischer There are two typos on the next-to-lowest line on the bottle: "E" -> "A" and "L" -> "M".

@marmot -- But this isn't honey liquor. At least, it isn't the right color.

@barbarabeeton I see, not only typos but also a color error. Better keep your claws away from this stuff. ;-)

hellas