plc-code transpiler limitations

Status (2026-05-29): all limitations below are fixed upstream in siemens-plc-tools (branch fix/transpiler-limitations). ruckig-scl now depends on siemens-plc-tools/packages/plc-code and no longer needs any of the workarounds. This file is kept as a historical record of the issues found while porting Ruckig to SCL and how they were resolved.

The SCL code itself was always 100% valid TIA Portal syntax — the limitations only affected the Python test bench. They have now been fixed in the transpiler, so the idiomatic SCL forms are used directly throughout ruckig-scl.

What was fixed

#	Symptom in the test bench	Root cause	Fix
1	Inter-FC call "IsFiniteLreal"(x := #v) in an expression became a string literal with :=→==	named-block calls were only handled in statement position; FC return value not captured	ExpressionTranslator extracts the call as a protected placeholder; call_named_block returns the FUNCTION value
2	"dbRuckigConst".RESULT_X stayed a string / int = 0	parser inserts spaces around the dot ("db" . X); DB pattern didn't match, and the enum-string collector grabbed < "db"	DB pattern tolerates whitespace; collector ignores quoted names followed by ./(; constant DBs auto-load from search paths
3	REGION Per-axis validation leaked - axis validation as code	_parse_region stopped the name at the first non-identifier token	name now consumes the whole line
5	Multi-line RHS in a REGION dropped continuation lines	_preprocess only joined a continuation ending in :=	also joins operator-led continuations
6	REGION Set 7 phase durations leaked 7 phase durations	same as #3 (digit)	same fix
7	Hex literal 16#8201 in code became 16 self.8201	# parsed as instance-var prefix before hex translation	hex translation runs first and tolerates parser-inserted spaces
9	An identifier starting with IF (e.g. "dbRuckigConst".IFACE_VELOCITY) is mis-lexed — the lexer reads IF as the keyword, producing invalid Python	The DB-constant pattern splits on ., leaving the bare IFACE_VELOCITY token which starts with the keyword IF	Workaround: use the literal integer value (e.g. 1) with an inline comment naming the constant (// IFACE_VELOCITY)
10	An inline comment on the same line as END_IF; (e.g. END_IF; // closes ...) silently drops the entire enclosing block	The control-flow depth counter matches END_IF only after rstrip of ; and spaces, not a trailing // comment, so the END_IF is not recognised and the block is discarded	Workaround: put the comment on its own line before or after END_IF;, never inline with it

(Items 4 and 8 in the original notes were conventions, not bugs.)

Idiomatic forms now used in ruckig-scl

• Shared constants live only in dbRuckigConst and are referenced as "dbRuckigConst".RESULT_WORKING, "dbRuckigConst".EPS_POSITION, etc. The test bench auto-loads the DB from src/data-blocks via the runtime's block search paths (see tests/conftest.py).
• Finite checks call the IsFiniteLreal sub-block: IF NOT "IsFiniteLreal"(x := #input.currentPosition[#i]) THEN ....
• REGION names use their natural form (Per-axis validation, Set 7 phase durations and jerks).
• Long expressions (e.g. the boundary-state integration) span multiple lines.

Test harness setup

tests/conftest.py exposes a make_harness factory with the standard search paths (src/blocks, src/data-types, src/data-blocks) so inter-FC calls, UDTs and the constant DB all resolve automatically.