> You are referring to the fact that "describe("integrate");" will hang
> the maxima-texmacs interface. That problem, ***and, to the best of my
> knowledge, all other ways to hang the maxima-texmacs interface*** have
> been eliminated by the new external interface code. I
Now that's good news! So I cancel my statement about the kernel interface and
claim the opposite.
> Personally, I would favor a communication protocol with higher-level
> two-way communication. (I envision something where the external program
> could request an operation or the value of a variable in a certain
> format. Maxima would return output with an indication of the output
> type, i.e., plain text, latex, mathml, postscript, binary data, etc.)
Maybe real two-way communication is not necessary. I like the way Axiom
handles its input: system commands, e.g. ")display properties SF", are kept
apart from mathematical input and don't increase the line number. A frontend
can always use system commands in the background without any notice of the
user. Having in mind a controller-document-view-model,
(maxima kernel)<n-----m>(notebook)<1-----k>(view),
The view can therefore provide all the system features without changing the
document in any way, e.g. increasing line numbering, command history.
> Besides, my biggest concern at this point is the Maxima kernel.
> Eliminating kernel bugs, mathematical and other, is the highest
> priority.
Of course, but maybe not the most fun...
Kind regards,
Michael