DECIDED 2026-08-01 — reject

User's call: reject, codebase is stale. [[wish-compile-gnu-pascal]] moved to rejected/.

decide: reject wish-compile-gnu-pascal, or keep it open?

• Type: decision — Track U. Scope call, no files, no gate.
• Status: open — filed 2026-07-20.
• Blocks: [[wish-compile-gnu-pascal]].
• Raised by: Track B queue sweep. The wish has carried a "rejection candidate (user call)" note since 2026-07-19 and was still ranked at prio 45 as available Track B work. Nobody had filed the actual question, so it sat in the queue looking like a task while waiting on a decision.

The question

Should GNU Pascal be a corpus target at all?

The case for rejecting

• GPC is not standalone-buildable. It is a GCC frontend, so "compile GPC" means building it inside a GCC tree, not compiling a self-contained project. That is a very different and much larger job than the wish implies.
• The sibling analysis already says so. idea-c-realworld-test-targets reaches the same conclusion independently and recommends p2c or tcc instead.
• tcc self-compile is already DONE, so the thing this wish would have proven — that the C frontend handles a real compiler codebase — is proven.
• GPC's Pascal runtime targets ISO 7185/10206. Our Pascal frontend targets the FPC/Delphi surface. Overlap is partial, so even the Track B half feeds the compat campaign less than Synapse, fgl or FPC itself already do.

The case for keeping it

Its ISO-dialect runtime is the one thing in the list we do not otherwise exercise — every current Pascal corpus target is FPC/Delphi-shaped. If ISO conformance ever becomes a goal, this is the ready-made corpus. That is a real argument, just not one that supports a prio-45 ranking today.

Recommendation

Reject, and note the ISO angle in idea-c-realworld-test-targets so it is not lost. Two independent sweeps have now reached the same conclusion; leaving it open a third time costs more attention than the option is worth.

Once decided

Reject -> move [[wish-compile-gnu-pascal]] to rejected/ with a pointer here. Keep -> drop its blocked-by edge and re-rank it honestly (well below 45, since nothing depends on it).

Log

• 2026-07-20 — Filed from the Track B sweep. Moving it to rejected/ IS the decision, which is why a Track B agent should not simply do it.