File size: 9,639 Bytes
9368cc4 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 | #!/usr/bin/env python3
"""Batch 6: more compound tasks.
Sub-defects are reused from tasks already validated individually, so each is
known to be findable and known to be covered by its own tests. Combining them
changes only how many must be found in one episode.
"""
import sys
from pathlib import Path
sys.path.insert(0, str(Path(__file__).parent))
from make_tasks import make, spec, CORPUS # noqa: E402
FLOW = "python/scheduler"
ROUTER = "typescript/router"
LEDGER = "python/ledger"
def tests_from(repo, task, *names):
base = CORPUS / repo / "tasks" / task / "tests"
return {n: (base / n).read_text() for n in names}
# ============================================== ledger, 3 defects
make(LEDGER, "audit-findings",
spec("python", "logic", 5, """
Three findings from the stock audit. They are unrelated; all three need fixing.
1. Valuation and cost of goods sold are wrong for any SKU received at more than
one unit cost. The ledger is documented as valuing inventory first-in,
first-out.
2. A SKU sitting exactly at its configured reorder point produces no
replenishment suggestion, and we ran out during the supplier lead time. The
reorder point is meant to be inclusive.
3. Catching a projection up a second time raises rather than applying only what
is new. A projection is supposed to be resumable, and catching up with
nothing new appended must apply zero events.
"""),
[("ledger/projections.py",
" oldest = lots[0]", " oldest = lots[-1]"),
("ledger/projections.py",
" lots.pop(0)", " lots.pop()"),
("ledger/policies.py",
" if level <= threshold:", " if level < threshold:"),
("ledger/store.py",
" return [e for e in self._events if e.seq > seq]",
" return [e for e in self._events if e.seq >= seq]")],
{**tests_from(LEDGER, "fifo-lifo", "test_valuation.py"),
**tests_from(LEDGER, "reorder-boundary", "test_reorder.py"),
**tests_from(LEDGER, "since-inclusive", "test_catch_up.py")})
# ============================================== flow, 3 defects
make(FLOW, "scheduler-regressions",
spec("python", "logic", 5, """
Three regressions reported against the scheduler this sprint. They have
separate causes; please fix all three.
1. Reading the graph can corrupt it. A reporting tool that collects the
dependents of each task into a set and annotates that set finds the graph's
own structure changed afterwards, and subsequent scheduling is wrong.
2. A task that fails to be admitted keeps hold of whatever it did manage to
take. A task needing two different pools, refused the second, leaves the
first held for the remainder of the run.
3. Runs containing a failure never finish. When a task in the middle of a chain
fails, its immediate dependents are skipped but everything below them stays
pending forever, so the run never reports itself complete.
"""),
[("flow/graph.py",
""" if task_id not in self._tasks:
raise UnknownTask(task_id)
return set(self._dependents.get(task_id, ()))""",
""" if task_id not in self._tasks:
raise UnknownTask(task_id)
return self._dependents.setdefault(task_id, set())"""),
("flow/resources.py",
""" requests = list(requests)
for request in requests:
if request.name not in self.capacity:
raise UnknownResource(request.name)
if self.free(request.name) < request.amount:
raise ResourceExhausted(request.name, request.amount,
self.free(request.name))
holding = self._held.setdefault(task_id, {})
for request in requests:
holding[request.name] = holding.get(request.name, 0) + request.amount""",
""" holding = self._held.setdefault(task_id, {})
for request in requests:
if request.name not in self.capacity:
raise UnknownResource(request.name)
if self.free(request.name) < request.amount:
raise ResourceExhausted(request.name, request.amount,
self.free(request.name))
holding[request.name] = holding.get(request.name, 0) + request.amount"""),
("flow/scheduler.py",
""" skipped: List[str] = []
for downstream in sorted(self.graph.descendants_of(task_id)):""",
""" skipped: List[str] = []
for downstream in sorted(self.graph.dependents_of(task_id)):""")],
{**tests_from(FLOW, "dependents-aliasing", "test_graph_isolation.py"),
**tests_from(FLOW, "acquire-not-atomic", "test_atomic_acquire.py"),
**tests_from(FLOW, "cascade-shallow", "test_cascade.py")})
# ============================================== router, 3 defects
make(ROUTER, "routing-regressions",
spec("typescript", "logic", 5, """
Three regressions in the router, from separate reports. All three need fixing.
1. Our static file route /assets/*path no longer serves anything in a
subdirectory, and when a single-segment request does match, the captured
parameter holds only that one segment. A wildcard should swallow the entire
remainder of the path.
2. Multi-select filters apply only the last value. `?tag=red&tag=blue` behaves
as though only blue were sent. Repeated keys should collect every value in
order.
3. Popular endpoints keep falling out of the route cache while paths hit once
at start-up survive indefinitely. Eviction is meant to remove whatever has
gone longest without being used.
"""),
[("src/trie.ts",
""" if (segment.kind === 'wildcard') {
params[segment.value] = parts.slice(i).join('/');
return params;
}""",
""" if (segment.kind === 'wildcard') {
if (i >= parts.length) return undefined;
params[segment.value] = parts[i];
continue;
}"""),
("src/query.ts",
""" if (out[key]) out[key].push(value);
else out[key] = [value];""",
""" out[key] = [value];"""),
("src/cache.ts",
""" const value = this.store.get(key) as V;
// reinsert so this key becomes the newest in iteration order
this.store.delete(key);
this.store.set(key, value);
this.hits++;
return value;""",
""" const value = this.store.get(key) as V;
this.hits++;
return value;""")],
{**tests_from(ROUTER, "wildcard-remainder", "wildcard.test.ts"),
**tests_from(ROUTER, "query-repeats-lost", "query.test.ts"),
**tests_from(ROUTER, "lru-recency-on-read", "cache.test.ts")})
# ============================================== flow, 2 defects
make(FLOW, "retry-subsystem-broken",
spec("python", "logic", 5, """
Two problems with retries, reported together.
1. A task that has failed and is waiting to be retried never becomes eligible
again, no matter how far the clock is advanced. The run simply stalls.
2. Separately, capacity is not returned when an attempt fails, so a pool drains
over the life of a run until nothing can be admitted.
"""),
[("flow/scheduler.py",
""" # RETRYING counts as a candidate: its backoff is enforced below by
# ready_at, and leaving it out would mean a retry never fires.
if state.get(task_id) not in (st.PENDING, st.READY, st.RETRYING):
continue""",
""" if state.get(task_id) not in (st.PENDING, st.READY):
continue"""),
("flow/scheduler.py",
""" waits out its backoff.
\"\"\"
self.pool.release(task_id)
task = self.graph.get(task_id)""",
""" waits out its backoff.
\"\"\"
task = self.graph.get(task_id)""")],
{**tests_from(FLOW, "retry-never-fires", "test_retry.py"),
**tests_from(FLOW, "resource-leak-on-retry", "test_resource_lifecycle.py")})
# ============================================== router, 2 defects
make(ROUTER, "negotiation-and-headers",
spec("typescript", "logic", 4, """
Two content-negotiation problems, likely separate causes.
1. Clients sending `Accept: text/html, application/json` with no explicit
qualities get JSON. When two acceptable types carry equal quality, the
client's own ordering is the tie-break.
2. Clients that spell the header `Content-Type` rather than `content-type` are
treated as though it were absent. HTTP field names are case-insensitive.
"""),
[("src/negotiate.ts",
""" const candidates = parseAccept(header).filter((c) => c.quality > 0);
let best: Candidate | undefined;
for (const candidate of candidates) {
if (!matchesAny(candidate.type, offered)) continue;
if (best === undefined) {
best = candidate;
continue;
}
if (candidate.quality > best.quality) best = candidate;
}""",
""" const candidates = parseAccept(header)
.filter((c) => c.quality > 0)
.sort((a, b) => b.quality - a.quality || a.type.localeCompare(b.type));
let best: Candidate | undefined;
for (const candidate of candidates) {
if (!matchesAny(candidate.type, offered)) continue;
if (best === undefined) best = candidate;
}"""),
("src/headers.ts",
""" private static key(name: string): string {
return name.toLowerCase();
}""",
""" private static key(name: string): string {
return name;
}""")],
{**tests_from(ROUTER, "negotiate-tiebreak", "negotiate.test.ts"),
**tests_from(ROUTER, "header-case-sensitive", "headers.test.ts")})
print("done")
|