How about providing LaTeX compatibility for Freemind? What I have in mind is for Freemind to be able to read correctly marked-up text, and to display it as mathematical formulae. (Also with error messages if the formatting is wrong).
-I
How about providing LaTeX compatibility for Freemind? What I have in mind is for Freemind to be able to read correctly marked-up text, and to display it as mathematical formulae. (Also with error messages if the formatting is wrong).
-I
That's an ambitious project.
Indeed - but it can be made less ambitious (at least for now) by only dealing with a subset of the LaTeX markup language. Starting with simple things like polynomials, sums, products and integrals is one way of approaching the problem.
After getting that to work, built in error checking would be nice to add next. Then how about various symbols, such as quantifiers and set membership.
-I
If it's any help, I use latex to produce the equations, view them with a dvi viewer and then use a capture utility ('import' under linux) to save the dvi equation as a jpeg. Then you can add the equation as a picture. Not as convenient as typesetting, but does the job with the existing software.
Yes, you can do that, but it's annoying. Rather than Freemind doing the job for you, you have to go through the process of typing the LaTeX source, compiling to dvi format, capturing the image, and then putting it into the FM database. Including LaTeX can be done, but awkwardly.
Take a look at how wiki's implement this:
http://www.freedevices.org/tiki-index.php?page=wikipluginlatex
Probably the simplest way to implement TeX or LaTeX would be to have a pair of delimiter characters in the text of certain nodes, e.g. $Latex Source$, and a button on the toolbar which when clicked starts a search for LaTeX source in the nodes of the mindmap. Text included inside the '$' characters would be extracted, fed to a latex compiler program, and replaced with an HTML image. Of course, the user would have to tell the system where to find the LaTeX modules, but that shouldn't be too difficult.
Also, there is the issue of making the search as fast as possible. For this, perhaps using a labelled graph approach would work? Each node would be labelled according to whether it or its subnodes had changed. If so, the search algorithm should then examine those subnodes recursively. If not, the nodes could be ignored, thus speeding the algorithm.
-I
Log in to post a comment.