How to access Atom via atomID?

Is there currently any function/method in the source code of MathLive that returns the atom (or atoms in the case of highlighting) for a specified atom ID? I can’t seem to find a way of accessing the Atom object corresponding to a particular span by the current implementation of ModelPrivate. What is the best way to go about achieving this?

[arnog]

The intent of MathLive is to replicate the layout that TeX implements. Certainly not the internal data structures which were conceived when the computing environments where extremely different. And as Neil points out, TeX was never meant to be interactive. I initially avoided introducing a tree structure (specifically a parent pointer in Atom) because TeX did not have it and does not need it for its layout (which is basically one pass without the need to backtrack to the ancestors). However, it is very convenient to have when you need to frequently traverse the data structure, in particular for editing. As both Neil and I have demonstrated it is possible to live without it (using a separate “path” structure), but the code is both simpler and easier to maintain when using an actual parent pointer.

[NSoiffer]

Back in the 90s when I wrote the math editor for Mathematica, I used shared nodes and relative positions to save space: 8mb RAM was a lot and Mathematica generates lots of shareable subexpressions. It did save *a lot *of space for larger outputs, but keeping track of the path (I used a depth first count to produce a selection range) added a ton of complexity to the code. The code has since been converted to use parent pointers and I suspect it got much shorter. I believe the cursor/selection is now represented by simple pointers to the first/last char node that is selected.

That’s my history 😃

Neil

On Sat, Jun 6, 2020 at 8:50 PM Arno Gourdol [e-mail] wrote:

Also, just to make sure I understood it right: so parent() and siblings() always refer to the parent and siblings (respectively) of the atom that is currently “adressed” via path?

Yes. The path represent the current insertion point (or head of the selection), and ancestor(), parent(), siblings() etc… refer to the “relatives” of the atom pointed at by path.

Since you mentioned it: how else would you implement it this time?

I would add a parent field to Atom so that a tree (graph) of atoms can easily be navigated and I would use a simple offset counting scheme to represent the caret.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/arnog/mathlive/issues/479#issuecomment-640152796, or unsubscribe https://github.com/notifications/unsubscribe-auth/AALZM3ASZ5APAVLHUA7EAL3RVMFAJANCNFSM4NV7HFKA .