Skip to content

7.4 billion tokens per month. 34 free LLM providers. 635 free model endpoints. All behind one /v1 endpoint, plus any custom OpenAI-compatible endpoint. Smart routing, automatic failover, encrypted keys. Personal experimentation only.

TypeScript

Loosen the compression timing caps off CI (#1030)

Tashfeen5bad9eb1fff1
diff --git a/server/src/__tests__/services/compression.test.ts b/server/src/__tests__/services/compression.test.ts
index 396f8d0d..ea7afab1 100644
--- a/server/src/__tests__/services/compression.test.ts
+++ b/server/src/__tests__/services/compression.test.ts
@@ -377,6 +377,14 @@ describe('compression engines', () => {
expect(savedChars).toBeGreaterThan(0);
});
+ // The wall-clock caps below are calibrated to GitHub's hosted runners, where
+ // they gate every push. Off CI they are only a courtesy, and a laptop
+ // running the suite inside Docker measures 2-3x slower than a runner (an
+ // outside verification on Node 22 saw a 10-11 ms median and 252-277 ms on
+ // the linearity payload), so the same numbers there only produce false
+ // failures. Loosen them where nothing depends on them; CI keeps them exact.
+ const slack = process.env.CI ? 1 : 3;
+
it('stays within the 100 KB synchronous performance budget', () => {
const payload = `${'plain status line without protected tokens\n'.repeat(2_500)}tail`;
const samples: number[] = [];
@@ -392,8 +400,8 @@ describe('compression engines', () => {
// 5 ms flaky (Node 22 measured 5.54–5.77 ms in otherwise-green CI runs).
// Keep a tight median guard while the p99/adversarial 25 ms caps below
// continue to catch material synchronous regressions.
- expect(p50).toBeLessThan(8);
- expect(p99).toBeLessThan(25);
+ expect(p50).toBeLessThan(8 * slack);
+ expect(p99).toBeLessThan(25 * slack);
const unbalancedJson = '['.repeat(100_000);
const adversarialStarted = performance.now();
@@ -401,7 +409,7 @@ describe('compression engines', () => {
config: config('lossless'),
recordStats: false,
});
- expect(performance.now() - adversarialStarted).toBeLessThan(25);
+ expect(performance.now() - adversarialStarted).toBeLessThan(25 * slack);
const manyBalancedArrays = '[]'.repeat(50_000);
const balancedStarted = performance.now();
@@ -409,7 +417,7 @@ describe('compression engines', () => {
config: config('lossless'),
recordStats: false,
});
- expect(performance.now() - balancedStarted).toBeLessThan(25);
+ expect(performance.now() - balancedStarted).toBeLessThan(25 * slack);
});
it('stays linear on protected-token-heavy adversarial payloads', () => {
@@ -427,7 +435,7 @@ describe('compression engines', () => {
], { config: config('standard'), recordStats: false });
const traceStarted = performance.now();
call(traceLines.join('\n'));
- expect(performance.now() - traceStarted).toBeLessThan(250);
+ expect(performance.now() - traceStarted).toBeLessThan(250 * slack);
// Distinct numeric literals are the highest-cardinality protected kind;
// the fidelity survival scan once rescanned the output per literal.
@@ -435,6 +443,6 @@ describe('compression engines', () => {
for (let i = 0; i < 12_000; i += 1) numberLines.push(`metric row ${i * 7 + 1_000_003} value ok`);
const numbersStarted = performance.now();
call(numberLines.join('\n'));
- expect(performance.now() - numbersStarted).toBeLessThan(250);
+ expect(performance.now() - numbersStarted).toBeLessThan(250 * slack);
});
});