docs: structures, enumerations and arrays in Python and C++ function blocks

[thiagoralves]

Adds two pages covering the types DOPE-584 unlocked for native blocks, and corrects what the existing pages got wrong. The docs still described the old world — base types and flat arrays only — while structures, enumerations, multi-dimensional arrays and arrays of structures all work now.

New pages

• Python Data Types — structures, enumerations, arrays (including the lower-bound trap), arrays of structures, strings, temporal types, function block instances
• C++ Data Types — the same ground on the C++ side, where almost every spelling differs

Everything here was run on hardware

Not written from reading the source. Each example was compiled and uploaded to an SLM-RP4 (Runtime v4) with openplc-cli, and the outputs read back through the debugger — 35 C++ patterns and 26 Python checks. Several of the findings contradict what I would have written from the code alone:

Finding	Why it matters
Member spelling is opposite. Python keeps the Variables Table spelling (mot.speed); C++ sees the compiler's uppercase (mot.SPEED).	Getting it wrong in Python is an AttributeError at runtime, not a build error. I hit this writing the examples.
Python lists use IEC indices. ARRAY [1..2] is a list of length 3 whose index 0 is None.	for row in grid: hands you a None first.
grid(i, j), not grid[i][j] for multi-dimensional arrays in C++.	[i][j] does not compile.
A scalar enum pin needs MODE m = mode; first. mode.get() returns a wrapper that will not cast to int.	An array element is already the raw enum and skips the step — the two differ.
A function block instance must be VAR to be driven from Python.	Declared as Input, all four pins still arrive and reads work, but writes never reach the PLC — a TON sits at zero with nothing to explain why. Verified both ways.
DATE is days; TIME/TOD/DT are nanoseconds.	Confirmed on device: D#1970-01-11 → 10, T#2s → 2000000000.

Also documents a failure mode that lies about itself: drop one of the four required imports and the block exits reporting "PLC runtime has stopped." while the PLC is plainly still running. The wrapper's liveness check calls os.kill, os was never imported, and the NameError is reported as a stopped runtime. Easy to hit, since the imports look unused.

Correction to an existing page

cpp-structure.md said ordering comparisons on STRING need .get() on both sides. They work directly on the variables; verified by compiling both forms.

Overlap with #14 — please read

#14 is open and touches two of the same files. It documents shared globals and function block instances, and its C++ instance section is correct — including that ton0(); is what makes the timer advance, which my hardware run confirms independently.

This branch is cut from main, so it does not include #14. Consequences:

• The new pages are new files and will not conflict.
• My edits to cpp-structure.md and python-restrictions.md are small but sit near docs: native function blocks — globals, instances, and the full IEC type surface #14's edits, so whichever merges second will need a trivial rebase.
• My C++ Data Types page has a short Function Block Instances section that overlaps docs: native function blocks — globals, instances, and the full IEC type surface #14's fuller treatment. If docs: native function blocks — globals, instances, and the full IEC type surface #14 merges first, that section should be trimmed to a link. Happy to do that, or to rebase this onto docs: native function blocks — globals, instances, and the full IEC type surface #14 instead — say which you prefer.

Not covered

• STRING[n] (a declared length) is still the transport's 126-character budget, so the docs state the cap rather than the declared length. Deferred to strucpp by DOPE-584.
• Negative array bounds (ARRAY [-10..10]) are legal IEC 61131-3 but rejected project-wide by both the parser and the variables table, so they are not documented as usable.

🤖 Generated with Claude Code

https://claude.ai/code/session_01UaSZK4LqFWtZpERcqnZ8uQ

[thiagoralves]

Folded into #14 — the two overlapped on cpp-structure.md and python-restrictions.md, and #14 already covered function block instances correctly (including that the C++ ton0() call is what makes a timer advance, which my hardware run confirmed independently).

Everything from this branch is now on #14 as 260504b, with the duplicate instance sections trimmed to links rather than kept twice. One PR, as asked.