From e0c6872cf40896c7be36b11dcc744620f10adf1d Mon Sep 17 00:00:00 2001 From: Norbert Preining Date: Mon, 2 Sep 2019 13:46:59 +0900 Subject: Initial commit --- .../sec-conclusions-know-limitations.tex | 38 ++++++++++++++++++++++ 1 file changed, 38 insertions(+) create mode 100644 support/latexindent/documentation/sec-conclusions-know-limitations.tex (limited to 'support/latexindent/documentation/sec-conclusions-know-limitations.tex') diff --git a/support/latexindent/documentation/sec-conclusions-know-limitations.tex b/support/latexindent/documentation/sec-conclusions-know-limitations.tex new file mode 100644 index 0000000000..1e865c9c84 --- /dev/null +++ b/support/latexindent/documentation/sec-conclusions-know-limitations.tex @@ -0,0 +1,38 @@ +% arara: pdflatex: {shell: yes, files: [latexindent]} +\section{Conclusions and known limitations}\label{sec:knownlimitations} + There are a number of known limitations of the script, and almost certainly quite a few + that are \emph{unknown}! + + The main limitation is to do with the alignment routine discussed on + \cpageref{yaml:lookforaligndelims}; for example, consider the file given in \cref{lst:matrix2}. + + \cmhlistingsfromfile{demonstrations/matrix2.tex}{\texttt{matrix2.tex}}{lst:matrix2} + + The default output is given in \cref{lst:matrix2-default}, and it is clear that the alignment + routine has not worked as hoped, but it is \emph{expected}. \cmhlistingsfromfile{demonstrations/matrix2-default.tex}{\texttt{matrix2.tex} default output}{lst:matrix2-default} + + The reason for the problem is that when \texttt{latexindent.pl} stores its code blocks + (see \vref{tab:code-blocks}) it uses replacement tokens. The alignment routine is using + the \emph{length of the replacement token} in its measuring -- I hope to be able to address this in the + future. + + There are other limitations to do with the multicolumn alignment routine (see + \vref{lst:tabular2-mod2}); in particular, when working with code blocks in which + multicolumn commands overlap, the algorithm can fail. + + Another limitation is to do with efficiency, particularly when the \texttt{-m} + switch is active, as this adds many checks and processes. The current implementation + relies upon finding and storing \emph{every} code block (see the discussion on + \cpageref{page:phases}); it is hoped that, in a future version, only + \emph{nested} code blocks will need to be stored in the `packing' phase, and + that this will improve the efficiency of the script. + + You can run \texttt{latexindent} on any file;% + \announce*{2019-07-13}*{ability to call latexindent on any file} if you don't specify an extension, then the extensions that you + specify in \lstinline[breaklines=true]!fileExtensionPreference! (see \vref{lst:fileExtensionPreference}) will be consulted. If you + find a case in which the script struggles, please feel free to report it at + \cite{latexindent-home}, and in the meantime, consider using a \texttt{noIndentBlock} (see \cpageref{lst:noIndentBlockdemo}). + + I hope that this script is useful to some; if you find an example where the script does + not behave as you think it should, the best way to contact me is to report an issue on + \cite{latexindent-home}; otherwise, feel free to find me on the \url{http://tex.stackexchange.com/users/6621/cmhughes}. -- cgit v1.2.3