@AlanMunn lambda calculus is used by linguists?

@Skillmon isn't it used by everyone?

@DavidCarlisle the lambda.sty is used by the skak package and it made the reading of the code quite challenging ;-). I'm not getting well along with such a notation.

@DavidCarlisle On the context list there is a question from a context user about how to access glyphs in the PUA ;-). "in a next beta i'll also support the P<original private> lookup"

@DavidCarlisle ooh

@Skillmon Those pesky linguists exploiting reductions and whatnot. :)

There's chocolate nearby

There was

:D

@UlrikeFischer I can see i'll have to subscribe in the end:-)

@PauloCereda beta reduction at work

@DavidCarlisle ;-). But it is interesting how topics emerges ...

@DavidCarlisle ooh :)

@DavidCarlisle From an end-user's perspective, being asked to choose between luatex and xetex is akin to being asked if you prefer Sachin Tendulkar to open the innings or come to the crease at two wickets down. Sure there are technical differences in each case, but no matter what, the job will be done, isn't it?

Random thoughts of a jobless Phd student waiting for viva

On the other hand, pdftex is like Rahul Dravid, reliable and everything, but not the most swashbuckling option out there :)

@Krishna not really. It depends what the job is. For some people the main point is using unicode and OTF fonts in which case they are comparable, but for others the main point of luatex is that they can use lua instead of tex for programming, and in that case xetex looks just like classic tex and not at all like luatex to them.

That's why I said end-user

Quack

@Krishna end (tex) users also program (look at questions on this site) or more to the point they use packages that have been programmed by someone. The majority of luatex-specific packages can't be ported to xetex in any reasonable way because they use lua.

Fair enough

@Krishna xetex's original design aim was to be using system supplied font libraries (originally the mac OS font libraries, now fontconfig/harfbuzz) the main design aim for luatex is to provide programmatic access to tex's internal structures. Unicode access is somewhat secondary.

Yes. By Jonathan Kew, I know

yes. I am aware of the differences ..... But if you are not using lua specific code, and merely want access to system fonts, then it doesn't make much difference for Latin characters

This is an oversimplification and I understand that, but for an end user who is an experimentalist for whom, tex is merely a tool for writing down the research, but who is fed up of the font constraints due to inflexible tex font metrics, then these both are equally viable options

@Krishna Don't say such things to e.g. Karl Berry ;)

@Krishna The danger with system fonts is they are not the same on every machine ...

@Krishna that's quite a tautology. If you don't use any of the features that are different then it naturally doesn't matter. But as soon as you care about microtype, or some other scripts, or want to use some graph drawing library, or ... then it matters.

@Krishna I might be able to agree with that version of your statement but the categorisation is a bit forced. A more natural statement might be "For users who just need the features common to xetex and luatex, it doesn't matter which they use"

oops @UlrikeFischer types faster than me!

@DavidCarlisle or eat faster than you ;-) I was just coming from lunch.

@UlrikeFischer your clock is an hour out.

@Krishna the vast majority of people who might be described by "an end user who is an experimentalist for whom, tex is merely a tool for writing down the research" do and should use pdflatex.

@DavidCarlisle ooh timelords

@DavidCarlisle I just read from your company portal that mathjax is shutting down their server

This is a bigger news than meets the eye, isn't it?

@Krishna that's old news (2 years ago?)

@Krishna it just means you have to use the cloudflare CDN URL instead of the old mathjax.org one

understood

Ok....That's what pandoc was always doing I guess

@Krishna well it must have changed as previously you had to use mathjax.org (or host it yourself of course)

@Krishna it was a pain for us at the time as we ship the html documentation as part of the product so the version on our website was a one line change, but we needed to contact all existing users and warn them to update their copies

I got functional shadings working in pgf-cmykshadings if anyone is interested to test (@samcarter)! All I can say is that at least PostScript is easier than BibTeX language… Interestingly, these don't work too well outside of Adobe Acrobat (both GhostScript and Cairo fail in some cases).

@Skillmon Yes, for example we typically model predicates as functions, so eat would be λxλy(x eat y) then apply the object so λxλy(x eat y)(apples) yields λx (x eats apples) etc. This is usually combined with a type theory, the simplest involving two types <e> for individuals and <t> for truth values. So eat is of type <e,<e,t>>, so when you apply the arguments you end up with the meaning of the sentence being a truth value, i.e., either true or false with respect to an assignment function.

@AlanMunn duck would have been a better example than apples

@DavidCarlisle Too late to edit. :)

@DavidCarlisle oh no

@DavidCarlisle duck eats apples?

@PauloCereda Wikipedia recommends Pato no tucupi as a highlight of Brazilian cuisine, do you recommend it?

@DavidCarlisle Oh no

@DavidCarlisle oh

@samcarter For this tex.stackexchange.com/questions/454646/… one thing that comes to my mind is to have some tree-like structure. You may put koalas and marmots under mammals, which means in this context that they could wear shoes and gloves, which ducks can't because they are not in this category. That is, at the highest (=tikzlings) level there are items which everyone can have, but then in the subfolders there are more specific items.

@marmot Would this work if I want to keep the individual packages independent of each other? At the moment my system is that each tikzlings is its own package and can be loaded alone. And the big tikzlings package just loads all the subpackages.

@samcarter without looking at the code I'd have expected the opposite structure with a tikzlings package containing the core which is required by each of the individual packages

@samcarter I'm rather certain that pgf has also some filter macros and that you can get the list of keys that a thing can handle from a larger list. If not switch to l3keys and then ask @JosephWright ;-).

@DavidCarlisle The tikzlings don't have much code in common. Each one has other abilities and uses (mostly) other keywords.

@UlrikeFischer :) or picture mode and bug @DavidCarlisle with my questions :)

@UlrikeFischer but filtering sounds good. I'll search on the main site if I can find something in this direction...

@samcarter I never get clashing keys with picture mode

@samcarter or the docu "key filtering".

@DavidCarlisle picture mode would avoid a lot of problems - but the main site would be very empty without all the tikz questions.

@UlrikeFischer found the chapter about it. On the first glance there might be something suitable in it. I'll have to leave now, but I'll read it carefully tonight. Thanks for the suggestion!

@UlrikeFischer pgf has filtering, of course I like our my approach

@JosephWright I prefer l3keys too. I do find the way of pgfkeys to define and use keys with one command confusing.

@UlrikeFischer I know :)

@UlrikeFischer If you remember, I asked during the development of keys3 about exactly that, and someone said they preferred the define/set split ;)

@UlrikeFischer pgfkeys is OOP, of course

@JosephWright ew

@PauloCereda l3keys is not :)

@JosephWright ooh

Any pandoc users here?

@AlanMunn user not really, but I'm able to run it.

@samcarter No, probably not. Yes, it is a tough question if you do not want to load everything. Some packages have their libraries or objects, like tcolorbox or tkz-euclide, but you also that many answers on the site are essentially: "You just forgot \usetkzobj{all}."

@UlrikeFischer I guess I'm at the same level. I use it for some things, but now I'd like to do a little customizing.

@UlrikeFischer For example, in the default latex template there is the following code:

$if(geometry)$
\usepackage[$for(geometry)$$geometry$$sep$,$endfor$]{geometry}
$endif$

This allows you to pass geometry options thus: -V geometry:margin=1in. My question is can I add similar code myself e.g. to load titlesec and pass it options?

Obviously it must be possible, but is it as simple as adding some code the default.latex or do I need to do more than that?

@AlanMunn Imho (if I remember it correctly from the time I made such template) you can simply add something similar $if(titlesec)$...

@UlrikeFischer That's what I thought too, but it doesn't seem to be working. It could be some silly mistake on my part of course.

I lately do get a strange warning quiet frequently:

pdfTeX warning (ext4): destination with the same identifier (name{table.
2.1}) has been already used, duplicate ignored

\AtBegShi@Output ...ipout \box \AtBeginShipoutBox
                                                  \fi \fi
l.427 \end{describemacro}

I'm getting this warning for all my floats at the start of the next page after they were outputted.

@AlanMunn it worked for my. I created a custom template my.latex, added

$if(ducks)$
\usepackage{tikzducks}
$endif$

and then run pandoc -s -f markdown -t latex -V ducks=1 --template=my.latex test.md -o pandoc.tex and had the tikzducks package in the result.

Without the variable it wasn't there.

@UlrikeFischer Hmm. It's not clear what I'm doing wrong then. I added

$if(titlesec)$
\usepackage[$for(titlesec)$$titlesec$$sep$,$endfor$]{titlesec}
$endif$

And then used -V titlesec:sf but no change.

@AlanMunn works for me. I get \usepackage[sf]{titlesec}. Are you sure your template is loaded? Change some of the fix text to test.

@UlrikeFischer Ah, I figured it out. I though pandoc would look automatically in ~/.pandoc for templates, but apparently not. When I change the name of the template I get an error ` Could not find data file /usr/local/Cellar/pandoc/2.2.3.2/share/x86_64-osx-ghc-8.4.3/pandoc-2.2.3.2/data/‌​templates/test.latex`

So pandoc is looking in a strange place for the template files, so my changes aren't being used.

@UlrikeFischer Oh well. Off to teach now, so I'll figure this out later. Thanks for testing.

Found the issue. hyperref was loaded too late (\AtBeginDocument) in conjunction with the scrreprt class.

@DavidCarlisle Your system is something akin to the RVPublisher demo that kaveh gave about 2 years back at a UK-TUG meeting?

@Krishna at work you mean? No we just edit xml files in emacs.

@Krishna were you at the meeting? have we met?

No....Although I'd have easily hopped on the train to Oxford if I had even a passive acquaintance

the presentation is here ....youtube.com/watch?v=hLRcd262J3k

@DavidCarlisle Kaveh showed an end-to-end publishing system driven entirely by XML

Sounds pretty awesome

@Krishna yes I know they were videod (I was there:-) I was just trying to recall who else was there that year.

@Krishna but that's an end user system so the xml is hidden in the background. We need to see the xml directly and edit the explicit markup.

@DavidCarlisle What is wrong with that system, wherein the XML gets enriched, but saves the human from the trouble of editing the XML

" We need to see the xml directly and edit the explicit markup. " Why do you need that?

Kaveh's system looks like a beefedup version of Pandoc suitable for publishing

And his demo was so perfect :)

Where can I try this out?

@Krishna you can't I think I think it's a service he sells to publishers so you'd have to submit to a journal that was using that....

Yes.....That's truly worth selling

Sounds awesome

@Krishna that's like saying why not write in Word and convert the result to latex via some process rather than directly editing the tex. for some uses that's fine but for others it isn't fine at all.

Right. Except that systems like Word and latex are human readable

@Krishna Word is nowadays saved as XML ...

XML isn't human readable. Isn't it a bit of a torture to make people edit a whole document in XML ?

Yes.....the docx I know....But stil lthe front end is reasonable

@Krishna xml is no less readable than tex if set up right

Is it?

@JosephWright yes but it's xml is to xml is like xii.tex is to tex

I have tried Oxygen Editor. Still looks pretty hard with all those tags

@Krishna yes a docx file (so anything after word 2012) is just a zip file of xml

Understand. I am really intrigued by your workflow. When I go to industry, I am sure they will make me edit in Word and use Excel!

I really want to carry forward my TeX and other skills to automate as much as possible through a *nix workfflow

@DavidCarlisle Can you please point me to some good resources out there for XML based document automation such as report generation etc.

@Krishna remember though that our XML isn't just a document for humans to read we generate hundreds of thousand of files of Fortran, C, java, python and C# code from it. You really don't want a wysiwyg editor for code

Right. Something like documentation of source code right? Like what minted does using the pygments library

@DavidCarlisle I am horrified at the thought of using Mundane tools like Excel and Word and all those point and click. But Industry captains swear by it. Alas!

@Krishna no it's the source of the code as well as the source of the documentation.

You mean like Knuth's literate programming?

@Krishna yes (well as like that as anything else)

Got it..... Something like an XML equivalent of a Jupyter notebook

@DavidCarlisle You said Emacs! Gosh.

Is that even an editor?

As opposed to some infinitely extensible lisp :)

@Krishna but far more extensive in the metadata for example look at the description of work here

nag.co.uk/numeric/fl/nagdoc_fl26.1/html/f08/f08bcf.html#work

@DavidCarlisle That documentation page isn't exactly nice looking. Sorry to say that

@Krishna the xml not only has the math markup for that, it also has the metadata to say that it is workspace that can be internally allocated in any interface not designed in the 1970s and the source that you can see is the same source of the C code for teh C interface which doesn't have a work parameter but internally does

WORK = NAG_ALLOC(MAX(1, ( (side == Nag_LeftSide) ? n*nb : m*nb )), double );

Understood

That line just being a restyling of the same XML input. We generate similar in fortran, python or whatever is needed. a wysiwyg editor for the source is the last thing we need.

Got it

Where can I learn some XML automation for TeX-based report generation

@Krishna don't ask me about wsiwyg I have spent an entire career avoiding it:-)

Sorry...was a typo

I hate WSYWIG

I ment XML based automation that reads data from a textual source (such as a csv file) and spits out a standardised report

The PDF of the report could be based upon TeX

@Krishna go and work for Hans:-) but actually tex input is very hard to process, journals do it (and often convert to xml) but it's much easier to control the process if you input in xml.

How do we achieve this? For example, it is useful for a university to typeset marksheets/transcripts of all their candidates

@Krishna ah if you are not taking arbitrary tex source then it's a lot easier of course

Right.......XML in the centre of your workflow like how Kaveh did it

Yes......is there some free resource or some textbook you could point me to?

@Krishna not really also it depends what you want to do, you could program it all in tex, (cf the datatool package) but many other systems have "latex output" where the programming is in python or lua or matlab or whatever and all the iteration over the csv data done in a "normal" programming environment that just writes out a tex file for back end pdf generation. Both ways have merits but the skill sets required are rather different of course

@DavidCarlisle I am aware of the datatool package. What are these other systems you are referring to?

@Krishna Joachim gave a nice talk at the TUG meeting in Germany about his system which was used for printing some ridiculous number of statements of (If i remember correctly) electricity bills so very simple tabular data but lots of them, so he had some system using plain tex using dvi as pdftex was too slow just turning out tens of thousands of very small documents all being printed via tex with no human intervention

Cool

Is it available somewhere?

@Krishna only if you work for some German electricity company:-)

Hmm....This is speeddata publisher, isn't it?

@Krishna lots of systems have tex output, sphynx for example there is speedata, or pandoc for that matter

@samcarter Thanks for this comment! ;-)

Understood.

@Krishna no, that's from Patrick. He is doing catalogues and things like this with it. It is xml based and use lua to print.

@marmot Ropes can be very useful, for example for things like i.stack.imgur.com/VEq3o.gif :)

@UlrikeFischer The documentation of speeddata is sparse

Do you think a speaker with only English knowledge can deploy it in production for say automated generation of reports from a csv file?

@samcarter these are coils, though. ;-) But, yes, I agree. At a given point, when I have more time, I want to try to combine them with the loops library by LoopSpace.

@samcarter @PauloCereda knows of good uses for ropes

@marmot I was just thinking, that one could recreate your jumping coil marmot with a jumping rope tikzling - therefore it would be really bad if the code for the ropes would be lost. And in contrast to many of the other draw-this-image questions, this one at least mentions the topic in the title, so it will be searchable. So I think it is good to keep it on the site.

@marmot It might also be useful to draw a plait of hair for the tikzlings

@DavidCarlisle To quote @PauloCereda "Ooh no!"

@samcarter Yes. @DavidCarlisle will be jealous ;-)

@marmot Will be difficult for him to do draw the same in picture mode :)

I think the answer is no but: Is it somehow possible to redirect the output of \showhyphens into another file than .log?

@Skillmon no

@samcarter I still have not found out what the purpose of closing a question that has an (even accepted) answer is. I always thought that one may close it if it is a duplicate, or if there are good reasons why the community does not want to provide an answer. But if there is also an answer....

@samcarter I was more thinking of this conversation ;-)

@DavidCarlisle Doesn't work -- dimension to large errors. ;-)

@DavidCarlisle but parsing the full log file seems unreasonable to reintroduce hyphenation into the argument of ulem :(

@Skillmon what's the actual question?

@marmot There is one understandable scenario to close a question that already has an answer: the question was flagged before the answer answer was posted but for people in the review queue, who see it after the answer, it is sometimes hard to notice if the questions has an answer. (if I remember correctly it depends on which queue it is in if the answer is shown greyed out or not at all)

@Skillmon you can set the text in a box of with 1pt to get the hyphenation points and re-constritute, but then you have a sequence of boxes not character tokens so it depends what you want to do....

@DavidCarlisle using ulem with automatic hyphenation (because soul seems too unstable). Or is there a solution I missed out to underline stuff line breakable and hopefully with working hyphenation.

@marmot Oh yes, that is certainly true!

@Skillmon why do you want to underline stuff:-) you could do as I suggest above or these days it's probably as easy to just use some \tikzmark like markers in the text and then go back and draw in the underline between the marks

@DavidCarlisle I did this, but you have to assume perfectly well behaving line skips to get it right for contents spanning multiple lines.

@DavidCarlisle is this the approach soul is taking?

@Skillmon er probably haven't looked for 20 years, hang on...

@Skillmon from a quick look over the code I suspect so note the \lastbox loop in \def\SOUL@analyze{{%

@Skillmon tikzmarks? no.

@UlrikeFischer you got that wrong, take a look to which that is an answer...

@Skillmon that is it's using the classic \lastbox iteration not tikz

@DavidCarlisle I know from what you wrote. I wasn't suspecting it using TikZ (why does everyone think so... :)

@Skillmon some people think tex==tikz

@DavidCarlisle not really. And it tends to be slow for something so "easy" like underlining.

@Skillmon yes if tikz wasn't loaded for other reasons you could just use \pdfsavepos and a primitive \hrule

@DavidCarlisle how to detect the end of line in \raggedright with this approach? Seems not worth it. Do you have any experience in parsing the log-file back in? I expect it to be really slow...

@Skillmon it would be quicker to use perl:-)

@Skillmon has to be said that doing all this in luatex would be so much easier

@DavidCarlisle yes, it would. But then I'd force a specific engine on others.

@DavidCarlisle But this means I've to force yet another external script onto others (and my perl isn't that good)

@Skillmon yes but one day something will snap, it's 4 times harder to maintain latex on {p,lua,pdf,xe}tex than just one of them and at some point enough features will only be available on one of them that the cost of multi-engine support will get too high

@DavidCarlisle and I guess LuaTeX will make the race eventually. But up till now pdfTeX isn't dead (who uses pTeX?).

@Skillmon ほとんどの日本のTeXユーザー

@DavidCarlisle :)

@DavidCarlisle hoton tono nihon no TeX yuuzaa :)

@Skillmon sie alle benutzen xml

@DavidCarlisle Why should anyone use that evil format? :)

@Skillmon it was very cool and trendy for a few weeks in 1998

@DavidCarlisle the compromise of human and machine readability by making it unreadable to both :)

@Skillmon of course it's easy to read it if you use a proper editor

@DavidCarlisle if a data format needs tools to be viewed upon it shouldn't be considered human readable.

@Skillmon all humans should read things in emacs (I've done that more or less every day since 1987 and it's not done me any harm)

@DavidCarlisle are you sure about that? Sometimes I get the feeling that you mistake operating systems for editors. Doesn't seem like you're still at full mental health :)

@samcarter Yes, that's true. I actually find the selection of what is shown when you go through the queue not optimal (and I never understood what the "Suggested edits" queue really is or what I am asked to do there, so I let it be).