@DavidCarlisle In the past I have found error messages less then useful/informative.

Though at the moment the only thing that immediately comes to mind is "there is no line here to end". Which is actually relatively useful. If I recall, that means there is a blank line (or possibly a return character) where there shouldn't be.

Have you seen the latest metapost question in the site?

@FaheemMitha well actually it's more literal than that, it means you have used \\ to end a line when there is no line started at this point. so you would get it after a blank line but also at the start of a parbox or anywhere else where tex isn't in horizontal mode.

@DavidCarlisle How does TeX define "no line started at this point"? A newline char immediately followed by another newline char with no tokens in between?

@PhelypeOleinik ;-) You shouldn't draw any conclusions unless you have the silver draw tag badge. ;-)

@PauloCereda tex.stackexchange.com/questions/486137/… Perhaps you should put a notice on github that the project was moved on gitlab (@HenriMenke)

@FaheemMitha no, as I said above it is nothing to do with newline characters, it is simply the use of \\ in vertical mode.

@CarLaTeX @HenriMenke ouch, deeply sorry! I was going to create a mirror, so the two repos are sync'ed, but there days are taking their toll on me...

@PauloCereda set up a bug tracker then leave the code until you have more time (say in 30 years) works for me....

@DavidCarlisle ooh

> TurboTax Uses Dark Patterns To Trick You Into Paying To File Your Taxes

news.slashdot.org/story/19/04/22/2139215/…

@barbarabeeton @AlanMunn ^^

@UlrikeFischer is what this says abut miktex/dvips (that it supports png directly) true? texfaq.org/FAQ-dvipsgraphics (if so dvips.def ought to support that, and it would be good to back-port it to texlive...)

@DavidCarlisle no imho this is no longer true. github.com/MiKTeX/miktex/issues/117#issuecomment-393584568

@UlrikeFischer ah, thanks

@DavidCarlisle quack

@PauloCereda have you seen this ?

@DavidCarlisle oh no

@PauloCereda Have you seen this ?

@UlrikeFischer As for the AR bug, it seems that one non-interactive link dummy per page is sufficient. If it is commented out, the link area to ctan.org remains active when the OCG is hidden in the Layers navigation panel of Acrobat Reader. In Evince it remains active in any case. The test input:

\documentclass{article}

\usepackage{ocgx2}

\makeatletter
\let\ocgbaseInsertOC\ocgbase@insert@oc
\makeatother

\ExplSyntaxOn
\let\pbsPdfLink\pbs_pdflink:nn
\ExplSyntaxOff

\begin{document}

non-interactive dummy:
\leavevmode\pdfstartlink user {
  /Ff 65537/FT/Btn/Subtype/Widget
}\phantom{dummy}\pdfendlink$\leftarrow$\\[3ex]

built-in:
\begin{ocg}{builtin}{builtin}{on}
  \leavevmode%
  \pdfstartlink user {
    /Subtype/Link/A<</S/URI/URI (ctan.org)>>;
    \ocgbaseInsertOC
  }\strut CTAN\pdfendlink

All other PDF viewers than AR I know of also fail to hide the active link area together with the PDF layer.

@AlexG Curious effect, there is nothing to see and the mouse doesn't change but if one click into the place the link is executed ;-) But if one dummy per page is enough to prevent this we should smuggle it in with some atbegshi code and not add it to every link @JosephWright.

@UlrikeFischer Yes, it's weird, indeed. But at least it can be worked around in AR by means of this trick. For the other viewers nothing helps. They don't acknowledge the /OC entry in the annotation dictionary at all. Perhaps, filing a bug report/feature request to Evince maintainers (or rather Poppler?) would be worthwhile.

@PauloCereda -- interesting. it's still possible to file on paper, where the only cost is postage and time to prepare the forms. that works -- we got our refund in near record time.

@barbarabeeton ooh

oops I just validated some xml and got 693919 lines saying error: element ^zz not allowed, who could have added that many <zz/> elements?

@DavidCarlisle your life-time job?

@UlrikeFischer yep:-)

I just tried to clip a box with l3draw and it doesn't work ;-(. @JosephWright There is something wrong in the docu (\driver_draw_box:Nnnnnnn doesn't exist). Beside this I simply don't understand the matrix values (and how to use the command). I see no box, and get messages about pdfTeX warning: pdflatex.exe: Misplaced \pdfrestore by (0sp, 11606383sp). (But I have to go now ...).

@DavidCarlisle ooh

@Skillmon ^^

@PauloCereda Coniglio alla cacciatora, very good!

@egreg oh no

Why didn't this answer get deleted given the review was all recommend deletion? tex.stackexchange.com/questions/486199/…

@StrongBad 28 flags remaining.

@UlrikeFischer Huh? Did you use the draw interface?

@UlrikeFischer I've fixed a typo, but as far as I know the code does work (test files pass)

@UlrikeFischer Probably do it by using a global bool; if true, output a widget and set to false, with the reset at shipout. This way, the first link on each page has the fix. It's right on top of a second annotation, not randomly at the start of the page

@UlrikeFischer Only issue occurs I guess if the page is broken during a link which was not creating a fix, and the next page has no links ...

@UlrikeFischer ^^^

\documentclass{article}
\usepackage{expl3}
\ExplSyntaxOn
\bool_new:N \g__driver_pdf_AR_fix_bool
\bool_gset_true:N \g__driver_pdf_AR_fix_bool
\cs_set_protected:Npn \__driver_pdf_AR_fix:
  {
    \bool_if:NT \g__driver_pdf_AR_fix_bool
      {
        \tex_pdfannot:D
          width  ~ 3pt ~ height ~ 3pt ~ depth ~ 0pt
          { \c__driver_pdf_AR_fix_tl }
        \bool_gset_false:N \g__driver_pdf_AR_fix_bool
       }
  }
\ExplSyntaxOff

\usepackage{atbegshi}
\AtBeginShipoutInit
\ExplSyntaxOn
\AtBeginShipout