MacPas conditional directives are ignored, so both arms compile

program mp;
{$setc TARGET_X := 1}
{$ifc TARGET_X = 0}
  {$error this arm must not compile}
{$elsec}
{$endc}
begin writeln('mp ok'); end.

fpc -Mmacpas compiles this and prints mp ok. pxx warns twice about unknown directives and then reports the error inside the arm it should never have entered.

Population

Directive words in fpc 3.2.2's own sources that pxx does not recognise, by occurrence — this family is the whole top of that list:

Concentrated in the MacOS bindings (univint), which is also the code most likely to be ported. Nothing else in that census exceeds single digits once the string-literal artefacts are removed — FPC builds directive text by concatenation in places ('{$ifde'+'f ...'), which a raw grep reports as a directive word ifde; those are not real and are not counted here.

Why "it warns now" is not enough

Compiling both arms is a wrong-program outcome, not a diagnostic gap. It happens to fail loudly when the dead arm contains something invalid (as above), and that is luck: two arms that are each independently valid — the common case for a {$ifc} selecting between two constant definitions — compile into a duplicate definition or a silently wrong constant.

Why this is parked and not rejected

Owner decision, 2026-09-05: "we target general fpc mode and even delphi mode, for now that is enough, our goal is not to support every pascal language on this planet, especially not if it's not in active use." Explicitly not never — which is why this is rainy-day/ (real, intended, deferred) and not rejected/ (the report is wrong) or low-prio/ (no plan, no claim). Nothing in the measurement below is retracted. What is retracted is the implication that 31549 occurrences ranks it.

The ranking defect this ticket carried

The population table is a census of fpc's corpus, and a ticket sourced from a corpus census inherits that corpus's priorities rather than ours. FPC ships univint because FPC targets macOS. So the number is precise, was checked carefully — the string-literal artefacts really were removed — and measures text we have no intention of compiling. Precision on the wrong quantity reads as authority. No program anyone wanted to build hit this; it was found by sweeping for directive words, not by attempting a target.

The same sweep also produced a genuine finding ({$A n} reported as a typo: ordinary Pascal, real code), so the instrument is not worthless — it cannot rank. A census ticket with no program we want to compile behind it needs a demand line or a lower prio.

The fork, if it is ever picked up

Either implement the family (it is {$ifdef}/{$if} machinery under different spellings, and {$setc}/{$definec} are {$define} with a value, which PasDefineValue already carries), or make an unhandled conditional directive a hard error rather than a warning — the same reasoning that made an unfindable include a hard error in bug-pascal-include-search-silent-miss. Ignoring a conditional is categorically different from ignoring a switch: a switch changes how code is compiled, a conditional changes which code is.

Found by censusing fpc's corpus for directive words this compiler calls unknown, after the unknown-directive warning landed — the census that also found {$A n} being reported as a typo. See the ranking note above before quoting the table.

Measured again 2026-09-05, with the mode line the ticket's repro omits

The repro above has no {$MODE MACPAS}, and real MacPas source does. Adding it changes nothing — same two warnings, same error from inside the arm that should never have been entered — because pxx's {$MODE} handler is DelphiMode := CaseEqual(name, 'delphi') and everything else falls into one bucket. That is the front door, and it is the half that is NOT parked -- it is fixed: {$MODE MACPAS} now errors, naming this ticket's measurement as the reason, so the repro above refuses at the mode line. What stays parked is IMPLEMENTING the conditional family. The two are independent: the front door stops us building the wrong program, and does not make us able to build the right one.