[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Octave-bug-tracker] [bug #35912] Implement a TeX interpreter for Octave
From: |
Burkart Lingner |
Subject: |
[Octave-bug-tracker] [bug #35912] Implement a TeX interpreter for Octave |
Date: |
Wed, 21 Mar 2012 17:01:09 +0000 |
User-agent: |
Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:11.0) Gecko/20100101 Firefox/11.0 |
Follow-up Comment #2, bug #35912 (project octave):
I wasn't aware that all TeX parsing takes place in gnuplot, so yes, I agree
that these bugs should be reported with gnuplot instead. However, I do have a
few remarks.
* This bug (#35912) doesn't concern TeX. Can you please change the title back
to "Wrong display of carets and underscores for
interpreter=='none'" (or something similar) and make one of the other,
TeX-related bugs the TeX master bug?
* While I generally do agree that bugs in the TeX parser should be fixed in
the TeX parser itself, i.e. in gnuplot, I don't think this applies to all the
bugs I submitted yesterday. Take for example bug #35920: gnuplot actually does
the right thing here in expecting ampersands and tildes to be escaped as "&"
and "~", respectively. This is compliant with standard TeX. Unfortunately
MATLAB's TeX parser has some quirks that make it incompatible with standard
TeX. Among those quirks is that it treats un-escaped ampersands and tildes as
printable characters. Now while Octave probably aims to recreate MATLAB's
brand of TeX, I believe gnuplot would rather implemenent standard TeX than
MATLAB TeX since it is used not only as a plotting tool for Octave. Therefore
I think Octave is responsible for transcribing MATLAB TeX to proper TeX in
order to pass that to gnuplot.
** You mentioned that you'd appreciate help in writing a (MATLAB) TeX
interpreter for Octave. I actually did recently write a conversion routine
from MATLAB TeX to regular TeX for the project matlab2tikz which you can find
at https://github.com/nschloe/matlab2tikz. What it does is take a
MATLAB-compatible TeX string as input and convert it to a string that can be
compiled with (La)TeX. Assuming that gnuplot interprets proper TeX correctly,
such an approach and some of matlab2tikz's code could be used in Octave as a
MATLAB/regular TeX compatibility layer for graphics_toolkit 'gnuplot', too.
* My understanding is that there is no TeX parser for graphics_toolkit 'fltk'
in Octave right now (except for the one at
http://octave.1599824.n4.nabble.com/TeX-interpreter-for-FLTK-backend-td3038693.html
which wasn't incorporated into Octave). This means that Octave users must use
graphics_toolkit 'gnuplot' to properly display TeX strings. If this output
then deviates from that which you see in MATLAB, personally I would regard it
as a MATLAB incompatiblity bug and not a feature request. Then again I don't
follow Octave development closely enough to know how you handle cases like
these.
_______________________________________________________
Reply to this item at:
<http://savannah.gnu.org/bugs/?35912>
_______________________________________________
Message sent via/by Savannah
http://savannah.gnu.org/