Sounds like your feedback could be really useful here!
I thought the big idea behind canvas was to avoid programming altogether through XML files...
Sort of, the primary idea is to make fully custom 2D drawing possible - without requiring any C++ or rebuilding.
What Tom mentioned is having a Nasal wrapper to parse the old HUD XML format in order to dynamically create a proper "canvas".
This could later on be used to extend the format and support additional modes.
The Nasal OOP is fairly well documented (see the wiki), and I'd volunteer to carefully explain all the details, and I'm sure Tom would also not mind helping you out if necessary
I must admit that I was raised in the times of low performance 8-bit computing and I am still reluctant to use an interpreted scripting language in a real time simulator.
fair enough, and you are right - however, keep in mind that Nasal is only just the "vehicle" here: The core canvas system is entirely written in C++ and works through the property tree ALONE. Nasal is only really used to set up things (initialization) and change a bunch of properties. Performance-critical stuff is usually not done by Nasal. And Tom has been really responsive to optimize things as required and move them to C++ space.
I want to use SVG files to set up some very basic symbols (the HUD ladder, the compass tape ... you name it) and then animate. That cannot be made easily generic I guess
There are plans to support this anyhow. And SVG files are already supported. Optimizing this is on the todo list. Animations were discussed for different needs, too. So probably, the canvas will get a handful of additional primitives/modes for animations, that can then be used for such needs. See the wiki for details.
when I was working for Rockwell-Collins, we used a tool to draw vector EFIS symbols that would then be animated by a specific API in the display unit... We are getting close to the real thing !
That's basically also how A661 works, which inspired parts of the canvas system - so the workflow should certainly be possible to support eventually.