TeX, LaTeX and Friends

12:08 AM

@samcarter_waiting_for_summer @StefanKottwitz davidcarlisle.github.io/latexcgi/test-sk#none

6:21 AM

@ComFreek Seems perfectly reasonable to me. I regularly read lists of things that begin at the highest number and end at number one. Like, if you need the person reading to keep scrolling to the end to find out which thing is number one. :p

Stefan Kottwitz

8:07 AM

@DavidCarlisle Very good, thanks! I like the local text features (keeping the language on site and update the .js file) and the regexps for norun/noedit etc. are nice!

David Carlisle

8:17 AM

@StefanKottwitz I used a different class name "button" seemed too generic and may clash with classes used by side builders like jekyll, so llbutton but otherwise should be fairly compatible with the version you had.

Stefan Kottwitz

@DavidCarlisle Would you like to extend the none regexp to match on typical TeX logfile start messages? That would have a good immediate effect on all the existing forum posts.

@DavidCarlisle I guess I like the none more than noedit because having the editor is the same look and feel and syntax highlighting, and line numbers help in pointing to (non-runnable) logfile lines.

David Carlisle

@StefanKottwitz I did wonder about allowing other first lines eg #!/bin/bash to trigger none without needing a !TEX comment. I'll have a look later, should be doable.

Paulo Cereda

@DavidCarlisle oh

@samcarter_waiting_for_summer ooh

@DavidCarlisle oh no

Stefan Kottwitz

@DavidCarlisle The first digits of \pi in the noedit regexp sounds a funny way to match TeX log files :-)

David Carlisle

@StefanKottwitz we could say the default is none if there is no \ or !TEX on the first non-white line. The main thing that would stop is the trivial snippet hello world which currently gets wrapped in a article class fragment. Or as you say we could specifically look for tex logs. But I am tempted to be more generic (eg for bibtex files etc)

Stefan Kottwitz

8:30 AM

There was no need for limiting the line numbers of the ace editor on learnlatex.org, but in general it would be good to have a nice limit (before a scrollbar comes) as code can be long

@DavidCarlisle Yes, it's good to think about that.

David Carlisle

@StefanKottwitz it is limited but generous you willl see it does editor.setOption("maxLines",100); pick a number :-)

AlexG

@UlrikeFischer Thank you, I will try this.

Stefan Kottwitz

@DavidCarlisle ah I see :-)

Ulrike Fischer

@AlexG wait a moment, I'm just finishing a better example ...

Plergux

@PauloCereda In the spirit of Duek... Queck! XD

Paulo Cereda

8:33 AM

@Plergux ooh :)

Stefan Kottwitz

@DavidCarlisle that's what my wife said about my credit card

David Carlisle

@StefanKottwitz 50 ?

@StefanKottwitz she trusts you with a credit card?

Stefan Kottwitz

@DavidCarlisle as long as I'm behaving well.

Ulrike Fischer

@AlexG I think the following is the right way (I had forgotten that I did add the option to create GoTo directly with \pdfannot_link, one doesn't have to use the special command.)

\RequirePackage{pdfmanagement-testphase}
\DeclareDocumentMetadata{uncompress}

\documentclass{article}
\usepackage{hyperref}

\hypersetup{colorlinks}

\begin{document}
\section{section}\label{blub}
\ref{blub}

\ExplSyntaxOn
\pdfdict_new:n   {l_alexg_action_setocgstate_dict}
\pdfdict_put:nnn {l_alexg_action_setocgstate_dict}{Type}{/Action}
\pdfdict_put:nnn {l_alexg_action_setocgstate_dict}{S}{/SetOCGState}

\group_begin:
  % local options:
  %\pdfdict_put:nnn {l_alexg_action_setocgstate_dict}{}{}

Joseph Wright

8:56 AM

@UlrikeFischer ???

Ulrike Fischer

@JosephWright The backend code adds /S/GoTo hard coded:

\cs_new_protected:Npn \@@_backend_link_begin_goto:nnw #1#2
  { \@@_backend_link_begin:nw { #1 /Subtype /Link /A << /S /GoTo /D ( #2 ) >> } }

Joseph Wright

@UlrikeFischer Ah, right: I think we are still very much experimenting here - the idea from memory was that the backend here is explicitly for 'GoTo' links :)

Ulrike Fischer

That means you can't change it to some other type. But this is okay, it is for the special case and you can generate a general link if you don't want it. I was yesterday only to tired to see this.

Joseph Wright

@DavidCarlisle, @StefanKottwitz How long before people notice that the one place you've not made TeXLive.net available is ... here

Ulrike Fischer

@JosephWright I already asked @DavidCarlisle to give me a script to add it ...

Joseph Wright

9:03 AM

@UlrikeFischer That sounds about right, yes: quickly reminding myself, pdfTeX/LuaTeX have a decided attribute that takes the goto name keyword, which is why we wanted an abstracted-but-dedicated backend function

@UlrikeFischer Yes, but I meant a meta post asking to have it as a general feature: at that stage, we might need a server upgrade! (@DavidCarlisle, @StefanKottwitz)

samcarter_waiting_for_summer

@DavidCarlisle Brilliant! Thanks a lot!

Ulrike Fischer

@JosephWright yes, I think it is fine. The main advantage of this function is that it complains if the destination doesn't exist, so normally you want to use it for goto links.

David Carlisle

@JosephWright there are long standing feature requests on the main meta for similar things (or even a simple copy button to avoid having to scroll to copy a code block) but they are regularly turned down by theose on power

samcarter_waiting_for_summer

Oh, github is funny! In notification mails about issues it masks email addresses with ***, so
\makeatletter
\def\beamer@framenotesbegin{% at beginning of slide
\gdef\beamer@noteitems{}%
\gdef\beamer@notes{{}}% used to be totally empty.
}
\makeatother
becomes
> \makeatletter
> ***@***.***{% at beginning of slide
> ***@***.***{}%
> ***@***.***{{}}% used to be totally empty.
> }
> \makeatother

They could at least check for a .

Paulo Cereda

@samcarter_waiting_for_summer ooh

David Carlisle

9:10 AM

@samcarter_waiting_for_summer we had that recently. If the text is authored on the site and marked as a code block it doesn't do that but if it came in from mail then you can't use the markdown code synatx...

Paulo Cereda

@DavidCarlisle ooh

TUG 2021 WEBSITE IS LIVE tug.org/tug2021

5

samcarter_waiting_for_summer

@PauloCereda Oh, wow! This site looks nice!

Skillmon

@PauloCereda thanks, Mr. News-Duck!

Paulo Cereda

@Skillmon <3

@samcarter_waiting_for_summer there's ducks in it :)

And Bärs

David Carlisle

@PauloCereda see the starred messages, Joseph told us ages ago.

Paulo Cereda

9:20 AM

And a duck disguised as a Englishbloke

samcarter_waiting_for_summer

OHHH, there is going to be a rare book tour!

Paulo Cereda

@DavidCarlisle oh

@DavidCarlisle what has a CFP ever done for us? :)

David Carlisle

@PauloCereda b...

Paulo Cereda

@DavidCarlisle oh no

samcarter_waiting_for_summer

How to decide with so many interesting workshops?

Paulo Cereda

9:22 AM

@samcarter_waiting_for_summer "hey guys see this rare book here /drops book and it breaks oopsie"

samcarter_waiting_for_summer

@PauloCereda :)

Skillmon

@PauloCereda I bring you the fif... /drops stone tablet TEN commandments!

Paulo Cereda

@Skillmon LOL exactly

samcarter_waiting_for_summer

@PauloCereda Wasn't there some project which recreates Elementa Geometriae in latex? We could just use this as replacement - nobody will notice the difference :)

Paulo Cereda

@samcarter_waiting_for_summer ooh there was a ConTeXt project of reproducing the Euclid bloke geometry book thingy

David Carlisle

9:25 AM

@samcarter_waiting_for_summer solo ve a los españoles.

Paulo Cereda

@DavidCarlisle CUIDADO, LLAMAS!

samcarter_waiting_for_summer

@DavidCarlisle Puedo decir tapas

Oh and the interviews are back! They were very interesting last year

Paulo Cereda

@samcarter_waiting_for_summer ooh

David Carlisle

@PauloCereda Deberías ser bueno en español, ya que vives cerca de Buenos Aires.

Paulo Cereda

People sent profile photos used in those biography sections in articles :)

samcarter_waiting_for_summer

9:29 AM

@PauloCereda Just found all the Bärs and ducks amongst the speakers, this will be a great event!

Paulo Cereda

@DavidCarlisle this hurts me in several layers. :)

David Carlisle

@PauloCereda I try to be helpful

Paulo Cereda

@samcarter_waiting_for_summer Joseph is disguised. :)

@DavidCarlisle quack <3

samcarter_waiting_for_summer

@PauloCereda Yeah, but we know in reality he's a duck

Paulo Cereda

@samcarter_waiting_for_summer twitter.com/paulocereda/status/1371756519692324870

David Carlisle

9:36 AM

@PauloCereda Conference Dinner

2

Paulo Cereda

@DavidCarlisle oh no

Ulrike Fischer

@PauloCereda but I seem to be back to be a Bär.

Paulo Cereda

@UlrikeFischer oh

samcarter_waiting_for_summer

@PauloCereda :D

Joseph Wright

@DavidCarlisle Yes: perhaps worth raising as we now have (to some extent) a workable plan that is parallel to the usual 'post an image' advice?

Paulo Cereda

9:44 AM

@JosephWright ooh a plan

David Carlisle

9:59 AM

@JosephWright but your comment about server is serious, a link on every code block from here might require some more substantial compute support at the back end.

Paulo Cereda

10:15 AM

@DavidCarlisle ooh

Stefan Kottwitz

10:44 AM

@DavidCarlisle They are a $70 million company that earns money from TeX.SE so they can easily setup their own server(s) ;-) with loadbalancer and your backend software and can also contribute to your CGI software

@DavidCarlisle Sure they can start trying our running server :-) on the long run I guess a commercial company doesn't need to rely on a DANTE paid server but we happily support.

Joseph Wright

@StefanKottwitz Well yes

@StefanKottwitz In all seriousness, should you and I make the case as mods? We could talk to the Powers and explore if it's at all viable

Stefan Kottwitz

@JosephWright Fine for me! It won't hurt to make the suggestion. I just don't see it as realistic as they reduced employees and try to keep sites the same way to be effective and low on maintainance effort/cost so I doubt a bit that they would make the effort to install a backend service for TeX.SE alone.

Joseph Wright

@StefanKottwitz Very true

Ulrike Fischer

11:16 AM

dvipdfmx-unsafe.cfg -- xelatex users won't like this. <-- @DavidCarlisle this will also affect the online runs, you perhaps will need to change the xelatex calls when you update to tl21.

David Carlisle

@UlrikeFischer yes I saw that :(

Stefan Kottwitz

1:25 PM

@David could we replace <br /> by \n before it goes into the editor? We lose line breaks in some situations.

David Carlisle

1:53 PM

@StefanKottwitz er we could but shouldn't need to: can you point to an example on site where it fails, I'll look later can't look now. Currently I take the .textContent of the code block so all markup has gone (but br should already be a newline) I could (like your old interface to overleaf) take the innerHTML but then you get all the markup and have to clean it up by hand, so if we can persuade textContent to do the right thing that would be easier but I'll look.....

David Carlisle

2:04 PM

@StefanKottwitz OK no need to point to an example, I can reproduce: br in a pre forces a linebreak but doesn't show up in textContent. Odd, OK I'll fix that.

Steven B. Segletes

@DavidCarlisle Hello, David. Based on your comment at tex.stackexchange.com/questions/207284/…, I tried to use miniltx to add color to my PlainTeX BONUS SOLUTION at tex.stackexchange.com/questions/44618/…. But it did not work, so I ended up using Joseph's code at tex.stackexchange.com/questions/207284/…. Did I do something wrong?

Paulo Cereda

ooh a bonus

David Carlisle

@StevenB.Segletes other than trying to use any of my code?

Steven B. Segletes

@DavidCarlisle Well, as they say..."it compiled"

David Carlisle

@StevenB.Segletes I can look later but possibly not now if something obvious dosn't pop up.

Steven B. Segletes

2:08 PM

@DavidCarlisle I never look a gift horse in the mouth. If you would prefer I ask it as a question, I could do that.

David Carlisle

@StevenB.Segletes miniltx is a mini version of latex and latex doesn't do ciolor either, you want \input color (which is a version of color.sty that also inputs miniltx)

Steven B. Segletes

@DavidCarlisle Son of a gun, that took care of it. I will revise my BONUS SOLUTION

Paulo Cereda

ooh a bonus

Stefan Kottwitz

@DavidCarlisle here is an example: latex.org/forum/viewtopic.php?f=30&t=34102

Steven B. Segletes

@PauloCereda Maybe it's time to step up to the plate, then. It is one of the greatest answers of all time, and is getting zero traction. YOU CAN CHANGE THAT!

@PauloCereda Thank you!

Paulo Cereda

2:18 PM

34

Q: Dynamically count and return number of words in a section

I am writing a proposal for which it is required to give the number of words of the main section in the proposal itself. I found this answer on how I can get a word count on a LaTeX document: Is there any way to do a correct word count of a LaTeX document? But this is only for entire documents....

Steven B. Segletes

@PauloCereda You went above and beyond! Many, many thanks.

Paulo Cereda

@StevenB.Segletes we ducks are very good at bonuses. :)

David Carlisle

2:52 PM

@StefanKottwitz yeo trhanks I have a plan possibly related to this :-)) developer.mozilla.org/en-US/docs/Web/API/HTMLElement/…

Stefan Kottwitz

@DavidCarlisle great!

David Carlisle

@StefanKottwitz don't let Ulrike know I looked at a manual.

Stefan Kottwitz

@DavidCarlisle a duck has seen it

David Carlisle

Nov 13 '17 at 16:38, by Paulo Cereda

@DavidCarlisle we ducks are good at keeping secrets

Stefan Kottwitz

@DavidCarlisle I asked @PauloCereda to keep it a secret, that ducks are good at keeping secrets.

samcarter_waiting_for_summer

3:10 PM

@PauloCereda albatross is mentioned in the journal of the French user group:

Paulo Cereda

@samcarter_waiting_for_summer ooh

@StefanKottwitz ooh secrets about secrets

Skillmon

4:13 PM

@PauloCereda secret agent duck? Bond, Duck Bond!

Paulo Cereda

@Skillmon vimeo.com/284348495

4

Skillmon

@PauloCereda :) A classic!

Steven B. Segletes

@PauloCereda LOL!!

Paulo Cereda

@Skillmon <3

@StevenB.Segletes <3

Bob

Im wondering about data types in interface3. How do clists differ from token lists with regards to how their data is stored. Is a clist just a wrapper and handler for what would otherwise just be a comma separated token list?

Joseph Wright

4:24 PM

@Bob Yes, a clist is really just a special case of tl - they are there as they are very common

@Bob The restriction on a clist is that we remove spaces and empty entries, as they can otherwise 'leak' from the user level - I'd see the clist module as one about convenience

Bob

Ok that is good to know.

I am currently using a combination of comma separated token lists stored inside a sequence \seq_put_right:Nn \my_sequence_variable {#3} where #3 could be <a, b, c, d>

and I am wondering if I use \seq_put_right:NV \ my_sequence_variable \my_clist_variable will it produce like-formated data inside the sequence

@JosephWright Hmm the spaces might be difference in this case. Is there a way to make a multi-dimensional sequence? This is why I am imbedding comma separated token lists into sequences. Perhaps there is a better solution?

Joseph Wright

@Bob We've not yet implemented a generic container data type

@Bob It will store exactly what it says in the clist

Bob

Is writing a clist to a token list (and vice versa) a fast action, or does it require parsing the lists?

barbara beeton

@Plergux -- The precursors to a probable volcanic eruption near Reykjavik have gotten insistent enough that they were reported here this morning on the news. I hope everyone is prepared, with strong-roofed shelters. The volcanologist who was answering questions said that any warning would provide only minutes. (I hope the volcano isn't expressing displeasure at your thesis.) Good luck.

Joseph Wright

@Bob ???

Skillmon

4:40 PM

@Bob What do you want to do, what's the use case? Why do you think you need this operation?

Bob

@JosephWright Let me rephrase. If writing a token list to a token list is order 1, and writing a clist to a clist is order 1, is writing a clist to a token list order 1 or order N?

@Skillmon The code I am writing is constantly shifting data back and forth from sequences to clists and to token lists, then doing data manipulation followed by a reverse order storage.

Skillmon

@Bob but why? Why do you need to transform data types this often?

Bob

So I am wondering about how to make sure I am not wasting compute time.

Skillmon

@Bob seems to me like you're doing something pretty strange, that should be implemented in another way so that you don't need all those conversions.

@Bob How do you transform your clists to tls?

(and vice versa)

Bob

@Skillmon I need to be able to parse multi-dementional data. I store user-entered comma separated token lists inside a sequence. Accessing this data involves pulling it out into a clist, then pulling each word out as a token list to parse it

Skillmon

4:44 PM

@Bob I don't quite get why you need to do the step seq -> clist in this setup. Seems like you could just as well manipulate elements of your seq directly.

@Bob Or pull tls from your seq directly.

Bob

\clist_set:Nx \tempclist {\seq_item:Nn \eventsraw {\monthcount}}
\int_step_inline:nnnn { 1 } { 1 } {\tempcount}{
\tl_set:Nx \templist {\clist_item:Nn \tempclist {##1}}

Skillmon

@Bob Yes, that's got much overhead. Why don't you use a \clist_map_... function?

@Bob \clist_item:Nn is comparatively slow.

Bob

how do map functions work?

Skillmon

@Bob just like your \int_step_inline:nnnn you could map over the elements of a seq with \seq_map_inline:Nn <seq-var> { <code> } (use #1 to access the contents of your seq elements inside of <code>)

@Bob or \clist_map_inline:Nn (works the same way, but \seq_map_inline:Nn should be faster, iirc)

Bob

And this would be faster than an \int_step_inline?

Skillmon

4:49 PM

@Bob this would be much faster than a \int_step_inline that does a \clist_item in each iteration.

Bob

@Skillmon What does the map function do internally if not a \clist_item equivilent?

@Skillmon Or is it because it doesnt have to restart counting every index call?

Skillmon

@Bob consider the following: \int_step_inline is a loop over $n$ elements. \clist_item:Nn has to do a \clist_count:N (loop over $n$ elements) and a loop over $i$ elements (with the first $n$ loop and the $i$ loop being inner loops inside of your \int_step_inline loop). All in all you'll get at least quadratic runtime.

Bob

@Skillmon Ok that makes sense. I did not realize indexing a clist was like a linked list rather than an array.

@Skillmon Is this the same for a sequence? Or are they order 1 indexable?

Skillmon

@Bob \clist_map on the other hand doesn't care for the number of elements in your clist, or the index of the currently processed element (assuming that you don't need it, even if you need it, stepping a counter in the body of the loop will be much faster). It just grabs the next element and applies your function/loop body to it, going forward through the list. Results in linear runtime.

Bob

@Skillmon Thank you for this. I will try to implement these mapping functions next.

Skillmon

4:55 PM

@Bob for sequences this works similar, I'd expect quadratic runtime for this as well (with an \int_step and \seq_item:Nn loop).

@Bob you're welcome. Oh, and applying map_inline functions is faster for seqs than for clists

@Bob the only data structures with constant access time are intarray and fparray. seq and clist both are at least linear.

@Bob but intarray and fparray both come with limitations (static size, for example, and obviously only numeric data)

Bob

@Skillmon That is too bad, but good to know!

Plergux

@barbarabeeton I noticed a little "jiggle" yesterday, otherwise I have not heard anything about any eruption :p My acquaintances who know about these are of the opinion that it would most likely be a "spill" rather than an explosion. So I'm not worried. :p

Steven B. Segletes

@Bob Since your goal is an L3 approach, I should just say nothing, as I know nothing. But if you were open to multidimensional list parsing using 2e techniques, the listofitems package may be of interest.

Joseph Wright

@Bob Everything else is in the end a macro under the hood, and that limits what we can do

Paulo Cereda

5:10 PM

ooh so there's a hood on top of a macro

I will shut up now

Plergux

@PauloCereda macro in da hood XD

Paulo Cereda

@Plergux ooh

Joseph Wright

@Bob Anything that works by expansion has to loop over 'all the content' unless we are lucky and TeX gives us some data structure. The only random access one is fontdimens, which are what are used for intarray, etc.

Skillmon

5:30 PM

@Bob if your stuff is really performance critical, you can define a wasteful data structure, which stores each element inside a separate token list (gives constant access time, but uses one control sequence per element plus 1 for each struct):

\documentclass[]{article}

\ExplSyntaxOn
\int_new:N \l__bobstruct_int
\cs_new_protected:Npn \bobstruct_new:N #1
  {
    \tl_new:N #1
    \tl_set:Nn #1 { \__bobstruct:n { \c_zero_int } }
  }
\cs_new_protected:Npn \bobstruct_set_from_clist:Nn
  {
    \__bobstruct_set_from_clist:fNn { \int_use:N \l__bobstruct_int }
  }
\cs_new_protected:Npn \__bobstruct_set_from_clist:nNn #1#2#3
  {
    \int_zero:N \l__bobstruct_int
    \clist_map_inline:nn {#3}
      {
        \int_incr:N \l__bobstruct_int
        \tl_set:cn

@Bob However, I do not advice you to use such a structure, sticking to the proper functions of expl3 (namely the map variants of your data type) should be preferred if viable!

barbara beeton

5:58 PM

@StevenB.Segletes '-- Just added that to thr "often referenced questions" list.

Steven B. Segletes

@barbarabeeton Thank you, Barbara. Now tell me, how do I see that list, which I did not know existed?

Joseph Wright

6:14 PM

@UlrikeFischer, @DavidCarlisle Fun with dvipdfmx ...

Phelype Oleinik

@StevenB.Segletes It's a big-list question in meta: tex.meta.stackexchange.com/a/7666/134574

barbara beeton

@StevenB.Segletes -- It's on meta: tex.meta.stackexchange.com/q/2419 Pretty useful, but one does have to know that it exists. There are several other members of this community who add to it when something good turns up. You're welcome to join the party.

Ulrike Fischer

@JosephWright yes saw it. Difficult decision.

Steven B. Segletes

@barbarabeeton Great! Thanks!

@PhelypeOleinik Thanks for the pointer!

Joseph Wright

6:35 PM

@UlrikeFischer I've made a few suggestions

Bob

7:01 PM

@Skillmon Woah its going to take me some time to work through that.

Skillmon

@Bob but keep in mind, using a seq should be preferred, especially if you don't really need constant access time (because you don't need to use one of the _item:Nn functions at all)

Alan Munn

7:12 PM

New website redesign. Not sure about the colour. amunn.github.io

David Carlisle

7:33 PM

@StefanKottwitz @samcarter_waiting_for_summer davidcarlisle.github.io/latexcgi/test2-sk#br

e38bd15 bib file example
1e3fe82 use none if first non white line not a tex comment and has no \ (eg log or bibtex files)
a735971 don't use latex mode for !TeX none case
c93b9b3 restrict editor to 40 lines
2ea1d82 use innerText to handle br and other visual elements

@StefanKottwitz do you have need for any plain tex examples in the forums? (it's currently not set up but it could be)

Stefan Kottwitz

@DavidCarlisle There's no need, I think, I hardly ever saw plain TeX in the forums.

@DavidCarlisle That looks very good! Your improvements will have immediate effect on the forum posts of many years, thanks to the automatic check so no new syntax enforced.

@DavidCarlisle where in the js file do you match bib and log files? I just don't see it at first glance (I see editor line limitation, latex for pstricks and others)

David Carlisle

@StefanKottwitz if a bibtex file starts with a tex comment it'll be seen as tex and get a submit button but you can live with that or add `% !TEX none" to turn t off, but mostly non tex things will automatiocally get none mode I think.

Stefan Kottwitz

I mean it works but I don't see the place where :-)

David Carlisle

@StefanKottwitz I just modifed the none regex const norunregex = /^\s*([/%#\*]+ *!TEX.*[^a-zA-Z]none *|[^% \t\\][^\\]*)\n/i;

@StefanKottwitz the bit after | says if the first non white line doesn't start with % and does not have a \ anywhere then assume it's not tex.

Stefan Kottwitz

7:49 PM

@DavidCarlisle ok, the \ that decides pro TeX is hard to see in /^\s*([/%#\*]+ *!TEX.*[^a-zA-Z]none *|[^% \t\\][^\\]*)\n/i :-) and I looked for matching the other way round and taking TeX by default

David Carlisle

@StefanKottwitz handling <br/> was simply one line: p[i].textContent=p[i].innerText;`

Stefan Kottwitz

@DavidCarlisle excellent! patching the forum code, in contrast, looked like a mess

barbara beeton

@AlanMunn -- It's all readable. What color did it used to be? (That's not properly grammatical, is it? I always have problems with "used to".)

Stefan Kottwitz

it would be a few lines but where to safely put and the next update breaks it

David Carlisle

@StefanKottwitz it worked on my manufactured tests with <br> and other markup, see if it copes with your forums:-)

@StefanKottwitz ?

Stefan Kottwitz

7:53 PM

@DavidCarlisle I mean when I looked through the forum code for custom bb code parsing it was hardly readable so I could not fix it there

David Carlisle

@StefanKottwitz ah:-)

Stefan Kottwitz

@DavidCarlisle yes, so your solution is perfect

David Carlisle

@StefanKottwitz if this survives a day or two at latex.org I may move some of it to the learnlatex version (not the fragment code but some of the more general code refactor), I'd like to keep them closer together just for my sanity and ease of maintenance...

Stefan Kottwitz

@DavidCarlisle funny: one line of simple text gets the LaTeX compile button, 2 lines don't: latex.org/forum/viewtopic.php?f=30&t=34102

David Carlisle

@StefanKottwitz I thought to myself at the time, will anyone notice:-) (I used \n in the regex to match end of first line which doesn't work if there is only one) I could fix that at expense of making the regex even more readable... I'll look at your example...

Stefan Kottwitz

8:03 PM

and a glitch on my side: The forums adds a small size intro top caption text like "Code: (can be compiled and edited, if valid LaTeX code and complete)" before the code to clearly tell users they can edit the code and compile it. I could shorten... But: can we have an intro caption text above the editor that only shows up when the buttons are added?

Because my text above non-compilable text snippets can get annoying :-)

@DavidCarlisle regexes don't need to be readable :-)

David Carlisle

@StefanKottwitz the one line thing is because your generated code doesn't have a newline before the closing </pre> as I say that is fixable. On your caption, it would perhaps be eaiest if the javascript inserted it at the same time it is inserting the button, currently you are adding it statically to the generated html?

Stefan Kottwitz

@DavidCarlisle right! at the same time (I did statically)

and yes, it would be good to fix the one line thing so it works without an \n at the single line

David Carlisle

@StefanKottwitz OK I can add the text to the lltexts array and inject a div with that in before the editor if it is non-empty, that way you could have captions or not.

Stefan Kottwitz

@DavidCarlisle perfect!

Bob

On the fourth line of this code segment, I do not see a prior definition of \__bobstruct:n
What does this notation mean?
\cs_new_protected:Npn \bobstruct_new:N #1
{
\tl_new:N #1%make new token list based upon parameter name input
\tl_set:Nn #1 { \__bobstruct:n { \c_zero_int } }
}

David Carlisle

8:10 PM

@StefanKottwitz see if it works first:-)

@StefanKottwitz just if there is a texlive.net button or also in the case the editor is inserted but no button?

Stefan Kottwitz

@DavidCarlisle only if there's a button

David Carlisle

@StefanKottwitz ok

Stefan Kottwitz

@DavidCarlisle btw. I guess editor.setOption("minLines",1); would be good for one line text (or did an issue with a single line occur?)

I'm fine with the editor for any text even if not compilable and editing won't make sense. Consistent appearance and syntax highlighting. If anybody edits without purpose it's an own choice :-)

David Carlisle

@StefanKottwitz I think I had issues with 1 when I first started experimenting with ace, someting about selecting text didn't work as I expected. You can try but I think it was 2 for a reason:-)

Stefan Kottwitz

that's why I'm fine with the editor for 1 line snippets and log files etc.-

@DavidCarlisle ok I try 1 at some time but not important

David Carlisle

8:26 PM

@StefanKottwitz as it is at the moment you never automatically get no editor you have to explicitly say !TEX noedit automatically you get the editor and the texllive button or not depending on the first line regex

Stefan Kottwitz

@DavidCarlisle yep fine

!TEX noedit
Dear publisher, please just print my book.

David Carlisle

@StefanKottwitz and if it's a one line book it now works the same whether or not it has a newline at the end: davidcarlisle.github.io/latexcgi/test2-sk#one-no-nl

Stefan Kottwitz

@DavidCarlisle looks good! latex.org/forum/viewtopic.php?f=30&t=34102

Sebastiano

8:42 PM

Good evening to all user connected in the chat....

@StefanKottwitz Hiiiiiiiiiiiii :-)

@DavidCarlisle Hi David also to you.

Sorry for this question. But I have seen this question tex.stackexchange.com/questions/587657/…

Is it possible that is it a duplicate? My English language is always very poor. Best regards everybody.

Stefan Kottwitz

@Sebastiano Hi! How are you?

Alan Munn

@barbarabeeton It used to be a mixture of green and purple. But no big color patches.

@barbarabeeton And I agree about 'used to'. I think you "should" have written "What color did it use to be?" but it's a very strange verb, so it's not really clear if that's linguistically reasonable, compared to what you wrote, "did it used to be" which is my intuition too.

barbara beeton

@AlanMunn -- Yeah, I think I changed it three times, and was about to reword it, but decided I might learn something. Or at least cause you a minor headache.

Alan Munn

@barbarabeeton The linguistically proper spelling should probably be 'usta'.

barbara beeton

@AlanMunn -- Love it!

Alan Munn

8:56 PM

@barbarabeeton But it's still a weird verb. Because it behaves like a modal verb in not having any tense inflection, but not like a modal verb in requiring 'do' in questions and negatives.

barbara beeton

@AlanMunn -- Only a problem when one writes it. Only in places like this will it cause me a problem. I don't try to write dialog for anything "serious".

Alan Munn

@barbarabeeton Indeed. But it's still a puzzling verb no matter how you spell it.

barbara beeton

@AlanMunn -- Maybe a topic for some poor confused student's master's thesis. It's suitably odd, while reasonably limited in scope.

David Carlisle

@StefanKottwitz with my usual subtle styling top captions: davidcarlisle.github.io/latexcgi/test2-sk#br

@StefanKottwitz you can use inline markup in the lltext string but more usefully perhaps the div has class lltopcaption so you can style it in the site css, if the lltexts entry is empty the caption div is not inserted at all

AlexG

@UlrikeFischer This looks very promising, thank you! Where can I find the documentation of \pdfannot_*, pdfdict_*?

In hyperref-generic too?

Ulrike Fischer

9:10 PM

@AlexG pdfannot is in l3pdfannot, pdfdict in l3pdfdict, you get a list of documentations with texdoc -l pdfmanagement

@AlexG do you know if ghostscript has a problem with filespec dictionaries?

AlexG

Ah, in different documents.

@UlrikeFischer Don't know. Which kind of problem?

Sebastiano

@StefanKottwitz Aside from the usual problems at school, my fatigue and with corruption running rampant even at school I'm fine. Fortunately I'm currently at home and it's windy here right now. When there are temperature changes or high winds my neck complains. With all my heart I am very tired, but when I connect with my friends at TeX.SE or answer questions or ask questions I relax a lot.

Ulrike Fischer

@AlexG I'm trying to change the GoToR type, it should no longer use /F (filename.pdf) but /F objref, where objref reference a filespec dictionary. It works perfectly with all engines but ps2pdf doesn't like it.

Sebastiano

@StefanKottwitz I hope you and your whole family are well. Here in Italy the armed forces and teachers are vaccinated with Astrazeneca. I am currently waiting for the vaccine to be improved and have no side effects. I respect the distances and indications of the scientific community.

Skillmon

@AlanMunn I'd prefer a tad darker (but I'm on the "light attracts bugs"-side of the pc user spectrum) and the blue with more saturation.

AlexG

9:21 PM

@UlrikeFischer Do you want it to point to a FileSpec dict that itself points to an embedded file stream?

David Carlisle

@StefanKottwitz grr your two-line error fragment gets the compile button as the end of line are marked as <br> so it sees \this on the first line. I'll fix...

Ulrike Fischer

@AlexG no, no embedded stream, only a Filespec dictionary.

AlexG

@UlrikeFischer Does GS emit an error msg?

Ulrike Fischer

@AlexG yes:

Error: /rangecheck in --pdfmark--
Operand stack:
   --nostringval--   _objdef   --nostringval--   Subtype   Link   Border   --nostringval--   H   I   C   --nostringval--   A   --dict:4/4(L)--   Rect   --nostringval--   ANN
Execution stack:
   %interp_exit   .runexec2   --nostringval--   pdfmark   --nostringval--   2   %stopped_push   --nostringval--   pdfmark   pdfmark   false   1   %stopped_push   2022   1   3   %oparray_pop   2021   1   3   %oparray_pop   2009   1   3   %oparray_pop   1865   1   3   %oparray_pop   --nostringval--   %errorexec_pop   .runexec2   --nostringval--   pdfmark   …

Sebastiano

@Rmano Hi and welcome :-)

AlexG

9:29 PM

@UlrikeFischer Try to create a minimal annotation with mark /_objdef{myname} /Rect [...] /ANN pdfmark And then add more entries to the annotation dict with mark {myname} /key/val ... /PUT pdfmark Maybe this works better.

I forgot: The /Subtype /<name> entry is mandatory for the minimal /ANN pdfmark.

Ulrike Fischer

@AlexG well I don't want a special dvips code path. So I need to understand why ghostscript doesn't like the object reference. It seems to complain about the {pdf.obj1} here:

SDict begin mark/_objdef{pdf.obj2}/Subtype /Link /Border [0 0 1] /H
/I /C [0.444706 0.456472 0.680002] /A <</Type /Action /S /GoToR /F
{pdf.obj1} /D [0/Fit] >>  pdf.rect/ANN pdfmark end

AlexG

@UlrikeFischer GS is sometimes a bit special. For example additional entries to an XObject dictionary created with BP and EP pdfmarks must be post-added with /PUT, while Distiller allows them to be added at once within the opening BP pdfmark.

Of course, for PUT to work, the annotation must be created with a name: /_objdef {...}

Ulrike Fischer

9:46 PM

@AlexG hm. Would it work if I create a separate object for the action dictionary?

AlexG

@UlrikeFischer Yes I think so, I am quite sure.

Ulrike Fischer

@AlexG well I will have to try, but it is a pity ;-(

Alan Munn

@Skillmon Thanks for the comment. Do you have a # (or #s) in mind?

Skillmon

@Bob it is defined a few lines down (in the 66th line, directly prior to the messages). \__bobstruct:n will just gobble the number representing the current element count of the structure and throw an expandable error of a misused bobstruct variable (all the mutator functions never expand \__bobstruct:n and it is also never needed to access the structure). It just serves as a safeguard if anyone would place a bobstruct variable in the input stream without a proper function.

AlexG

@UlrikeFischer Let me know whether it worked out.

Skillmon

9:51 PM

@AlanMunn no not really. Just my subjective impression (I prefer more saturated colours for such structural highlighting on web pages in general), and I really like dark web designs (the internet is a way too bright place for my eyes). But I guess since your site isn't about some hacking and programming intensive stuff, your target audience doesn't share that taste and the brighter design is better suited. Still a bit more saturation wouldn't hurt, imho.

@AlanMunn that is something I really like about the overleaf web page. It does use saturated (and well tempered, if that's the correct word) colours. Or the colour of the top bar on learnlatex, another example of good saturation in my eyes.

Alan Munn

@Skillmon Yes, I like the learnlatex colour. Not so wild about the Overleaf scheme though.

Skillmon

@AlanMunn not the general scheme, just the green :)

Alan Munn

@Skillmon True, but I don't like the green either. :)

Ulrike Fischer

@AlexG it seems to work, at least I didn't get an error:

      \pdf_object_unnamed_write:nx{dict}{\pdfdict_use:n {l_hyp/annot/A/GoToR}}
        \pdfannot_dict_put:nnx {link/GoToR}{A}{\pdf_object_ref_last:}

Rmano

@Sebastiano hi to you and... Good night ;-)

Skillmon

9:56 PM

@AlanMunn well, colour seems to be highly subjective, who would've guessed!

AlexG

@UlrikeFischer Hope it did make it in the PDF too. Did you check?

Ulrike Fischer

@AlexG yes, now the interesting test will be, how it handles non-ascii filenames (for which I'm doing all this ...)

AlexG

@UlrikeFischer Good luck!

Ulrike Fischer

@AlexG it worked. Adobe just opened grüße.pdf ;-)

AlexG

@UlrikeFischer Glad to hear this!

Ulrike Fischer

10:01 PM

@AlexG thanks for the tip with the two step process, I would never have thought of this ;-)

Skillmon

AlexG

@UlrikeFischer Yeah, it once took me quite a few trials and errors...

Skillmon

@AlanMunn ^^^ just a quick edit with Gimp, I'd probably go more in that direction colour wise (that's the same hue, but darker and more saturated).

@AlanMunn But you have to be careful, that's awfully close to Words default colour scheme... :)

@AlanMunn and of course, feel free to ignore my uneducated opinion. I don't consider colour design one of my strong points :)

Ulrike Fischer

@AlexG the escaping of the mime type is wrong ;-(. With dvips I get /Subtype/application#232Fpdf instead of /Subtype /application#2Fpdf in the pdf. How do one escape a slash in postscript?

Alan Munn

@Skillmon Ah, true. I hadn't thought of that. I toyed with a bright orange, but then thought it was a bit too bright.

AlexG

10:24 PM

@UlrikeFischer Don't know, right now :(, sorry.

Alan Munn

@Skillmon How's this:

AlexG

@UlrikeFischer I found it out. Use the cvn (convert to name) operator : /Subtype (application/pdf) cvn

Ulrike Fischer

@AlexG hm, this sounds as if we need another backend function. I need to think tomorrow about it.

AlexG

10:40 PM

@UlrikeFischer Maybe there is yet another way, but this one, cvn, is the most robust one, I guess.

Ulrike Fischer

10:54 PM

@AlexG it should be possible to adapt the convert function I use here to do this for dvips.

Alan Munn

@Skillmon Thanks a lot for your comments, they were really helpful.

Ulrike Fischer

@AlexG does it work for all unicode?

Transcript for

$TeX - LaTeX$ TeX, LaTeX and Friends