Bugs in asciidoc latex backend
==============================
Geoff Eddy
Benjamin Klum was unable to maintain the LaTeX backend beyond version
8.2.7, and as a consequence the `latex.conf` file ceased to work with
`asciidoc` sometime after this. For version 8.4.3 I made some changes
to the file to get it working again. This document summarises the
remaining issues which I have found with generating LaTeX and thence
PDF output with `asciidoc`, which is done as follows:
- `asciidoc --backend=latex --unsafe FILE.txt`
- `latex FILE.tex`
- `kpdf FILE.pdf`
Many of these were found by processing the `asciidoc.txt` file and
comparing the HTML output with the PDF.
== Footnotes
Priority:: Low.
Problem:: References to footnotes, and a sensible value for the
`[footnoteref-inlinemacro]` section, don't seem to be possible.
Cause:: LaTeX doesn't support footnoting much beyond creating a
footnote at a certain point in text and displaying the footnote
itself.
Solution:: Unclear. How important or necessary is this, anyway?
== Spurious text at starts of paragraphs
Priority:: Medium
Problem:: It is necessary to insert spurious text in paragraphs.
Cause:: This `asciidoc` input:
+
-------------------------------------------------------------------
Text
-------------------------------------------------------------------
+
generates this LaTeX code:
+
\begin{lstlisting}
SPURIOUS TEXTText\end{lstlisting}
+
which should be:
+
\begin{lstlisting}[]
Text\end{lstlisting}
Solution:: Find out a way to generate the correct LaTeX output as
above. The obvious solution, as explained in `latet.conf`, doesn't
work.
== Tables
Priority:: Rather high.
Problem:: Not all of the table configuration options can be passed
through to the LaTeX backend. In particular, I've had to assume that
all tables will be fifteen or fewer left-justified columns wide.
Cause:: The table models in LaTeX and HTML are too dissimilar for one
`asciidoc` specification to generate valid output in both formats by
simple regexp replacement. Related to this is the unfortunate fact
that `<COLGROUP>` and related tags aren't a required part of HTML4,
and some broswers (at least Firefox and Konqueror) don't implement
them.
Solution:: Perhaps table processing could be handled by a Python
plugin, which would read in a table spec and generate the appropriate
text?
== Unicode escaping
Priority:: Rather high, to me at least.
Problem:: The commented-out section in `latex.conf`, if uncommented,
converts `π` to `\unichar{960}`, which then causes LaTeX to
complain that the resulting command is unavailable in encoding T1. The
more common non-ASCII characters, such as those in `félicité` and
`świeca`, are handled properly, but some - such as the IPA characters
in the `tipa` package - are not.
Cause:: The encodings in the LaTeX output are wrong.
Solution:: Correct the encodings.
== Text colours
Priority:: Probably low
Problem:: The text colour options are not processed by LaTeX; for
example `[#ff0000]#Red text#` is not rendered in red.
Cause:: LaTeX and HTML represent RGB triads differently: HTML is happy
with `#ff8000`, but LaTeX needs `[rgb]{1,0.5,0}`.
Solution:: Provide some sort of internal RGB conversion mechanism
which can convert RGB triads to different representations.
== Text sizes
Priority:: Probably low
Problem:: The text size options are not processed by LaTeX:
`[,,1]#text#` is rendered in the same size as normal text.
Cause:: HTML size tags - `h1` through `h7` - are directly derivable
from the size number, whereas LaTeX has a series of descriptive words
(`HUGE` through `normalsize` to `scriptsize`).
Solution:: Provide a way to treat the number as an index into an
array.
== Background colour in paragraphs
Priority:: Medium
Problem:: If the `backgroundcolor` attribute is specified in the
`lstset` command, all paragraphs are displayed as black rectangles by
`kpdf`, `xpdf`, and `evince`, although `kdvi` has no problems. I've
had to remove the attribute, and so paragraphs all appear on white
backgrounds. The PDF viewers also complain:
Error (NNNN): Unknown operator 'rgb'
Cause:: Apparently a known bug in the output of `pdflatex`. Not a bug
in `asciidoc`.
Solution:: Wait until this bug is fixed?
