#2816 Significant reduction in processing LaTeX output in imaxima in new version

[Alex Grede]

The LaTeX output processing is taking seconds to complete each time which is making imaxima unusable. This was also commented on in imaxima LaTeX errors.

From what I can tell the function imaxima-dump-tex is run at start-up of imaxima. This runs the LaTeX processor using two files it generates and uses the output (mylatex.fmt) to significantly reduce the time subsequent LaTeX calls are made from imaxima-tex-to-dvi. Currently without any user options specified, this fails resulting in this error in mylatex.log:

! LaTeX Error: \boxed undefined.

l.9       \renewcommand{\boxed}
                               [1]{\fbox{\m@th$\displaystyle#1$}}

This likely started with [6dc101] to fix a conflict with users wanting to use the amsmath package (bug [#2718]). Two LaTeX commands defined in amsmath that conflicted were changed in imaxima-dump-tex from \newcommand to \renewcommand which fails if the command is not already defined. Adding \providecommand before \renewcommand will define the command if it is not already allowing \renewcommand to be run without failure.

This change also likely caused bug [#2741] fixed with [873f25]. This patch fixed the problem by adding back the parts that were missing when imaxima-dump-tex fails to produce mylatex.fmt.

The fix was rolling back [873f25], aside from the addition of \\usepackage{color} in imaxima-latex (around line 612) which may have been a different issue, and adding \providecommand where appropriate in imaxima-dump-tex to keep the fixes for amsmath (bug [#2718]).

[Robert Dodier]

Thanks for the detailed information. Do I understand correctly that the attached patch fixes the very-slow-TeX-processing bug?

[Alex Grede]

Yes it does

[Robert Dodier]

I have applied the patch as commit [acb216cc]. Closing this report. Thanks for the report, and the patch.