\DocumentMetadat{testphase=phase-III, ...} #1010
Replies: 9 comments 38 replies
-
|
the tagging of a bibliography with natbib looks more or less ok (it would probably be better if the author names were in the Lbl), but the list looses the hanging: with block-tagging
without the testphase code
|
Beta Was this translation helpful? Give feedback.
-
|
hyperref thoroughly breaks the tagging of the bibliography as it doesn't pick up the new A better fix is imho to change the kernel definitions so that they add targets. Then hyperref can leave the definition alone and only adjust the target names: |
Beta Was this translation helpful? Give feedback.
-
|
Hi, from \DocumentMetadata{testphase={phase-III,math}}
\documentclass{article}
\begin{document}
\begin{verbatim}
test \@setupverbinvisiblespace
\end{verbatim}
\end{document}and using dvi mode, there is a warning but then a fatal error with nobreakspace in verbatim |
Beta Was this translation helpful? Give feedback.
-
|
hm, yes. We will have to decide if we disable tagging in dvi mode or provide dummys for commands like \pdffakespace. |
Beta Was this translation helpful? Give feedback.
-
|
Hi, then problem with Console output: trace (line number is from original from which MWE was extracted). So there seems to be missing a |
Beta Was this translation helpful? Give feedback.
-
|
Yes, I missed a long. This has been resolved already in the github and will be corrected in the next dev-release (probably end of next week or so). |
Beta Was this translation helpful? Give feedback.
-
|
This with \DocumentMetadata{testphase={phase-III}}
\documentclass{article}
\usepackage{doc}% (v2 used in original source, same issue)
\begin{document}
\section{Documentation}
nothing here
\StopEventually{\end{document}\endinput}
\section{Implementation}
\MakePercentIgnore
% \begin{macro}{\mst@enc}
% \begin{macro}{\mst@fam}
% \begin{macro}{\mst@ser}
% \begin{macro}{\mst@opsh}
% \begin{macro}{\mst@bold}
% \begin{macro}{\mst@ltsh}
% Macros to store the font settings, each math version will store its
% own records.
% \begin{macrocode}
\def\mst@enc{\encodingdefault}
\def\mst@fam{\familydefault}
\def\mst@ser{\seriesdefault}
\def\mst@opsh{\shapedefault} %% will be default shape for operator names
\def\mst@bold{\bfdefault}
\def\mst@ltsh{\shapedefault} %% will be default shape for letters
% \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
\FinaleEnd of console output: log attached |
Beta Was this translation helpful? Give feedback.
-
|
hm, differently to standard LaTeX the new list code issues errors if trivlists are nested too deeply. @FrankMittelbach ? |
Beta Was this translation helpful? Give feedback.
-
doc is not yet tagging aware and it is messing around with grouping levels in nasty ways, so it doesn't surprise me at all that this is not going to work at the moment. We never claimed that anything other basic Lamport environments work (and even for those we aren't claiming that all work). So we can file this as "doc.sty" needs to made tagging aware but any further investigation is premature (unless it is shown that nested trivlists by themselves (without doc.sty around) show this behavior --- can't check, I'm traveling) |
Beta Was this translation helpful? Give feedback.
-
|
@FrankMittelbach it is not doc. Try this with and without the testphase code: |
Beta Was this translation helpful? Give feedback.
-
|
Beta Was this translation helpful? Give feedback.
-
|
Now that you say it @davidcarlisle yes that is the reason. My rationale back then was that given that we do not missuse trivlist for things like "center" etc trivlist is basically going out of business and unrestricted nesting is neither necessary nor it is happening in reality. With doc I presume it does happen because (as I said) doc messes around with applying trivlist in a strange way and I guess to the code it currently looks as if a sequence of "macrocode" environments is seen as if they are nested into each other. Now the decision to make trivlist a version of list (and not the other way around), could be revisited but I would first like to see any real life example (other than one with doc) that requires nesting them to any high level --- because in my opinion that doesn't really make sense at all. But of course doc would need to get an adjustment to make it tagging and avoid the trivlist issue as well. |
Beta Was this translation helpful? Give feedback.
-
|
Would it be possible to emit only once at end of run. I am currently experimenting with a document which generated about 5370 such warnings which occupy more than 16100 lines at end of log and in Terminal. |
Beta Was this translation helpful? Give feedback.
-
|
@u-fischer wrote: The Sect structure is around the section title and the text of the section. I thought we had we had decided not to use the PDF Sect type in this way. It is not really intended for such LaYTeX sections, etc. |
Beta Was this translation helpful? Give feedback.
-
Indeed, sorry for not having spotted it earlier but this is code from close to ten years ago. I was redefining both About:
I have now gotten rid of all tagging warning after a somewhat arduous bisecting which I wish I had been more clever to do instantly. Turns out I had at some other location in the document a direct usage of a \DocumentMetadata{testphase={phase-III}}
\documentclass{article}
\begin{document}
\section{foo}
\indent
\vbox{%
foo
bar
}
\end{document}Such a source generates many problems: After commenting out both the place where verbatim was hacked and the place where for whatever reason a |
Beta Was this translation helpful? Give feedback.
-
You need a \par at the end of the \vbox (or use LaTeX commands to avoid the problem ...). See ltpara-doc.pdf (end of page 2) why the missing par is a problem for the para hooks. |
Beta Was this translation helpful? Give feedback.
-
ah yes thanks; I knew about that but this is what happens when one goes down from 130 pages to 2 lines, one forgets about resolving and is focused only on localizing. |
Beta Was this translation helpful? Give feedback.
-
|
@u-fischer apologies but actually the original source is something like which I probably did to solve some placement problem. This has other problems... I strangly reported a situation after having removed the verbatim because due to my editor wrong syntax highlighting I thought the |
Beta Was this translation helpful? Give feedback.
-
|
In my test (which proceeds from mathastext.dtx, but modified to use |
Beta Was this translation helpful? Give feedback.
-
I will contribute a concrete example. Can you briefly explain what should be discussed here and what should go to tagpdf github? |
Beta Was this translation helpful? Give feedback.
-
|
only adding here for background that compilation time of my circa 130 pages document goes from about 2seconds to edit: the above with pdflatex. With lualatex it is rather 5 seconds without tagging and 3 minutes with tagging, again for only one pass. |
Beta Was this translation helpful? Give feedback.
-
The main point is not the github, but that real issues and bugs shouldn't be burried to deep inside a discussion. |
Beta Was this translation helpful? Give feedback.
-
well tagging does a lot. So getting slower is expected. Also the code currently concentrates on readability, not speed. But with lualatex at least it should be possible to activate tagging only at the final or the last two compilations, I need to think about options here. And naturally it is always possible that there is a specific bottleneck (e.g. a faulty access to a property) somewhere that one can correct, but for this a sensible example is needed. |
Beta Was this translation helpful? Give feedback.
-
Beta Was this translation helpful? Give feedback.
-
|
Warnings such as this one do not give the input line number. I don't know if this is possible, but without input line number I can not even attempt to try to understand what is going on. Perhaps the struct numbers provided help when viewing an uncompressed pdf, but I am not sure I do have access to tools allowing to build such pdf. Perhaps via online tool. |
Beta Was this translation helpful? Give feedback.
-
|
adding the line number is easy, and is imho a good idea. I changed the message in tagpdf. You can produce an uncompressed pdf simply by using the uncompress option in |
Beta Was this translation helpful? Give feedback.
-
|
FWIW: I have now finished the test to adapt my document to phase-III+math tagging, the main change have been to avoid fatal errors:
as reported earlier here and here. However I have not checked closely output yet to see further alterations to layout beyond the issue #74 with verbatim. (I see now some indentation problem with There are problems with After reactivating usage of scratcl class, tagpdf warnings re-surface but they all are the same I found in the log: but it occurs after the loading of latex-lab-footnotes.ltx. In this My document was not using extra KOMA syntax for usage of the sectioning commands so does not appear to be otherwise affected. |
Beta Was this translation helpful? Give feedback.
-
|
Where should I report this This was very tortuous to reduce to minimal example as it turned out that there are 3 different things I could do, completely unrelated one with the other, to get the problem go away. But the main thing is that the problem goes away if \DocumentMetadata{testphase={phase-III}}% used for enumitem emulation
\documentclass{article}
% uncomment this to fix problem
% \usepackage{enumitem}
\begin{document}
% original source is doc of mathastext and uses intensively
% mathastext. Now in that context the following line is
% present, and if commented, nothing else being changed,
% the problem does not show. But this is separated
% investigation so here we remove all mathastext thing
%\MTversion{normal}
% comment this to fix problem
\renewcommand{\familydefault}{\sfdefault}
$\bullet$ is bullet
\begin{itemize}[label=$\bullet$]
\item
is not bullet, why?
\end{itemize}
\end{document}outputs: |
Beta Was this translation helpful? Give feedback.
-
|
ah well, it appears the enumitem activation does not fix the problem created by |
Beta Was this translation helpful? Give feedback.
-
Well the source is the new block and list code (which you can load without activating tagging with The source of your problem is, that there is no label key yet (try |
Beta Was this translation helpful? Give feedback.
-
|
thanks for explanations; |
Beta Was this translation helpful? Give feedback.
-
|
well no my error is using |
Beta Was this translation helpful? Give feedback.
-
We are in the process of enabling phase-III results in the LaTeX Tagged PDF project, including automatic tagging of heading commands, various display block environments and mach more (basically the full set of commands and environments as discussed in Leslie Lamport's LaTeX manual, eventually).
Once that is available (which will be the case in a few days if you use one of the development formats, e.g.,
pdflatex-dev) this is a good place to provide feedback on the new functionality.Some further info (more is needed) is given in the draft version of
ltnews37.pdf.Beta Was this translation helpful? Give feedback.
All reactions