feat(local): add read-only failure-triage classifier over single_builds

Implement steps 1 + 2 of issue #115: turn recorded build failures into
triaged patch suggestions instead of hand-scraping Crow logs.

- add local/failing-builds-classify.R with pure, DB-free helpers to
  normalise error_text into a stable fingerprint and classify it against a
  seed signature set (removed tbb_stddef.h, RcppParallel bundled TBB,
  system libuv link leak), each carrying a fix tier, confidence, and an
  auto/human-only flag
- add local/failing-builds-report.R, a read-only entrypoint that queries
  single_builds WHERE error_occurred, groups by root cause, classifies
  each group, and prints (optionally emits JSON) a triaged report with a
  pre-filled registry.json entry for known, safe fix levers only
- keep novel source diffs and unknown signatures routed to human triage,
  per the issue guardrails; never write to the DB or the registry
- cover the helpers with local/tests/test-failing-builds-classify.R and
  document the workflow in local/patches/README.md
This commit is contained in:
Patrick Schratz 2026-07-14 14:05:53 +00:00
commit d9fb88f530
No known key found for this signature in database
GPG key ID: 62050D5BC68AB6DC

View file

@ -0,0 +1,87 @@
source(file.path("..", "failing-builds-classify.R"))
test_that("normalise_error collapses temp paths, versions, and package tokens", {
a <- normalise_error(
"In file /tmp/RtmpAb12/foo.c: RcppParallel 5.1.9 failed at 0xdeadbeef",
package = "RcppParallel"
)
expect_false(grepl("RtmpAb12", a))
expect_false(grepl("5\\.1\\.9", a))
expect_false(grepl("0xdeadbeef", a))
expect_true(grepl("<pkg>", a))
expect_true(grepl("<tmp>", a))
# Same root error across two versions collapses to one fingerprint.
e1 <- "StanHeaders 2.32.1: tbb/tbb_stddef.h: No such file or directory"
e2 <- "StanHeaders 2.33.0: tbb/tbb_stddef.h: No such file or directory"
expect_identical(
fingerprint_error(e1, "StanHeaders"),
fingerprint_error(e2, "StanHeaders")
)
})
test_that("normalise_error is safe on NA/empty input", {
expect_identical(normalise_error(NA_character_), "")
expect_identical(normalise_error(""), "")
expect_identical(fingerprint_error(NA_character_), "<empty>")
})
test_that("fingerprint_error picks the salient error line, not the last line", {
txt <- paste(
"* installing *source* package 'foo' ...",
"error: bar.h: No such file or directory",
"* removing '/tmp/lib/foo'",
sep = "\n"
)
fp <- fingerprint_error(txt, "foo")
expect_true(grepl("no such file", fp))
expect_false(grepl("removing", fp))
})
test_that("classify_error matches the removed TBB header signature (makevars, auto)", {
sig <- classify_error(
"fatal error: tbb/tbb_stddef.h: No such file or directory"
)
expect_identical(sig$id, "tbb-stddef-removed")
expect_identical(sig$tier, "makevars")
expect_true(sig$auto)
expect_true(sig$matched)
})
test_that("classify_error matches RcppParallel bundled-TBB failures", {
sig <- classify_error(
"Error: USE_TBB=Linux is not supported on this toolchain"
)
expect_identical(sig$id, "rcppparallel-bundled-tbb")
expect_identical(sig$tier, "patch")
expect_true(sig$auto)
})
test_that("system libuv link leak is classified but stays human-only (novel diff)", {
sig <- classify_error("cannot open shared object file: libuv.so.1")
expect_identical(sig$id, "system-libuv-link-leak")
expect_true(sig$matched)
expect_false(sig$auto)
})
test_that("unknown signatures are never guessed at", {
sig <- classify_error("segfault: memory not mapped at address")
expect_identical(sig$id, "unclassified")
expect_false(sig$matched)
expect_false(sig$auto)
expect_true(is.na(sig$tier))
})
test_that("propose_registry_entry fills a schema-valid entry for a known lever", {
sig <- classify_error("tbb/tbb_stddef.h: No such file")
json <- propose_registry_entry(sig, "StanHeaders", c("alpine", "ubuntu-2604"))
entry <- jsonlite::fromJSON(json, simplifyVector = FALSE)
expect_identical(entry$package, "StanHeaders")
expect_identical(entry$versions, "*")
expect_identical(entry$platforms, list("alpine", "ubuntu-2604"))
expect_identical(entry$makevars$CPPFLAGS, "-DTBB_INTERFACE_NEW")
expect_true(nzchar(entry$reason))
# No template -> no proposal (unclassified path).
expect_null(propose_registry_entry(classify_error("weird"), "x", "alpine"))
})