Dynamic databinding variable expressions

[Dakror]

Prompted by the submission of #547 here is my branch for fully dynamic expressions to access array indices in databinding addresses.

This is done by enabling a fast-path for the pre-existing fixed addresses, and a stack based index resolution of dynamic addresses.

I did not submit this PR earlier because i feared another CSS Variables incident with a giant PR destined to fail.

[mikke89]

Very nice, I like it!

Are there any limitations to this approach compared to #547? I'm curious whether it works on the left-hand side of data-value, data-checked, and data-event assignment expressions.

Does it properly report variable names of nested addresses? E.g. will the expression arrays.c[arrays.b[i1] - 19].val be re-evaluated if i1 is dirty?

[Dakror]

Indeed i needed to fix it tracking the embedded variables. But now it works.
Left-hand sides are not supported yet, the new instruction is read-only.

[mikke89]

Hey, did you get a chance to take a look at expressions on left-hand side/assignments?

I wonder if we still need something like #547, or if there is another way to approach that on top of this PR. Any thoughts?

[Dakror]

Left-hand side is definitely doable. Would you want to merge this first and go for dynamic assignment in a separate PR or all in one?

[mikke89]

I don't mind either way. I'll start reviewing this code now, and you can decide if you want to reuse this, or make a new PR. If you add everything here, it would be nice if you keep the commits clean, or do an interactive rebase at the end.

[Dakror]

If you're starting your review i'll hold of on adding assignment support, so i dont have too many headaches at once :D

[mikke89]

Overall, I very much like what you achieve with only changes within the data expression code. Well done! The feature seems very useful, so thanks for making the PR.

I have some suggestions here, most of which are relatively small.

[mikke89]

Nice work! Just a couple of small comments :)

Do you want me to merge this PR once these are addressed, and open a new PR for assignment support?

[mikke89]

Awesome work, thanks a lot! That's a very neat feature :) And I'm very happy with the implementation, and particularly that it could be implemented entirely in the parser, nice.

Only missing piece now is expression support in assignments.