math content in `picture` environment is lost

[mbertucci47]

With math tagging loaded, any math content in a picture environment is lost.

\DocumentMetadata
  {
    lang=en-US,
    pdfversion=2.0,
    pdfstandard=ua-2,
    testphase={phase-III,title,math,table,firstaid}
  }
\documentclass{article}
\begin{document}
\begin{picture}[alt=a test picture](5,5)
text$x$
\end{picture}
\end{document}

I first noticed this in the feynmf/feynmp example in #659.

[davidcarlisle]

The input should probably be \put(0,0){text$x$} (although the math still vanishes) text at the top level in a picture mode doesn't generate an error for historical reasons, but doesn't really do anything sensible)

[mbertucci47]

@davidcarlisle You're right, I was just trying to make the example as minimal as possible

[davidcarlisle]

@mbertucci47 sure no problem but then I wondered if that was part of the issue (but it isn't) (@josephwright will have a look...)

[josephwright]

Ah, tagging is not active inside pictures: likely one for @u-fischer

[josephwright]

~....\tag_socket_use:nn #1#2->\bool_if:NT \l__tag_active_socket_bool {\UseSocke
t {tagsupport/#1}{#2}}
#1<-math/inline/formula/begin
#2<-x

~.....\bool_if:NT #1->\if_bool:N #1\__prg_T_true:w \fi: \use_none:n 
#1<-\l__tag_active_socket_bool 
{\ifodd: (level 1) entered on line 11}
{false}
{\fi: \ifodd (level 1) entered on line 11}

~......\use_none:n #1->
#1<-\UseSocket {tagsupport/math/inline/formula/begin}{x}
{math shift character $}

[josephwright]

I guess we need to reset tagging status at the start of any boxes in picture mode.

[u-fischer]

it is not only picture mode. Math disappears always if tagging is suspended:

\DocumentMetadata
  {
    lang=en-US,
    pdfversion=2.0,
    pdfstandard=ua-2,
    testphase={phase-III,title,math,table,firstaid},
    uncompress
  }
\documentclass{article}
\begin{document}

xxx \SuspendTagging{test} $a=b$ \ResumeTagging{test} aaaa

\end{document}

[davidcarlisle]

@josephwright

I guess we need to reset tagging status at the start of any boxes in picture mode.

That's not clear: the outer picture env has [alt=a test picture] so I thought from a tagging perspective it ends there and is just a black box after that with the alt text filling in.

Don't we need to revert the math grabbing to normal if tagging is suspended?

[josephwright]

Don't we need to revert the math grabbing to normal if tagging is suspended?

We need to use the 'measuring/fake math/empty math' path at the end of grabbing - @u-fischer is on it :)

[u-fischer]

@josephwright but before I work on it I have to listen to an interesting talk

[car222222]

Are there not some larger questions here?

Concerning the tagging of the content of diagrams (as opposed to "pictures" in the non-technical sense: e.g. photos or paintings); Because such diagrams typically consist of many structural components, including for example: "text" of various types; very specific, often named, substructures (that are part of the semantics of the diagram); Noting that the semantic relationships between these substructures are distinct from, and more important than, any visual relationships.

At some stage, these will be coded by largely semantic markup (plus possibly some essential visual/layout hints).