Makefile: declare every target phony, and unhide integ-arm64 from help

[fpt]

Two small, independent Makefile fixes.

.PHONY listed only integ

Every other target was a real file target. A file or directory named test, run, fmt, lint or help in the repository root silently turns the matching target into a no-op — and it exits 0, so CI would go green without running anything:

$ make test
tests ran
$ touch test
$ make test
make: `test' is up to date.
$ echo $?
0

test is a plausible name for someone to create. All seven targets are now declared.

make help hid integ-arm64

The pattern was ^[a-zA-Z_-]+: — no digit class — so integ-arm64 carried a ## comment that make help never printed. Adding 0-9 fixes it, and it stays correct for any future target with a version or architecture in its name.

One comment, no behaviour change

A header saying what the Makefile is: convenience assuming GNU make and a POSIX shell, rather than the build. The build and tests are the Go toolchain alone, which is what lets the Windows CI job run go test ./... directly.

Verification

• make help now lists integ-arm64
• touch test && make test runs the suite instead of reporting "up to date"
• go test ./... passes against libduckdb 1.5.5
• golangci-lint run → 0 issues

Credit

The fixes and their rationale are Calvin Cheng's, from #38. That PR was closed in error — its author compared against their fork's main, where the changes had genuinely landed, rather than this repository's. Reapplied here as a single commit against the correct base.

Closes #38.

[github-actions]

🤖 AI Review (klein) — turn 1

The change correctly marks all documented Makefile convenience targets as phony, preventing root-level files from suppressing their recipes. It also expands the help-target grep pattern to include digits, so integ-arm64 is now listed without changing recipe behavior. I verified the surrounding Makefile: the declared target names match the actual targets, and the updated pattern matches the documented target names. I found no correctness, compatibility, or security issues in this change.

---

turn 1 (full review of ffcf2d5) · comments: total 0, active 0 (+0 new, −0 resolved) · verdict: approve