Where this was measured. A 12-core macOS host, load average 2.6 to 5.9, - * on commit 20c1094. The same four cells were also run under load and gave the same answer, but - * that run is not clean evidence: the load average climbed from 7 to 50 while the cells ran, and - * the old version ran last, at the top of that climb. Above about load 20 everything here is - * slow for reasons that have nothing to do with this seam. So read the claim as "measured near - * idle on a 12-core host", and nothing stronger. If this test fails on a smaller or busier - * machine, raise {@link #OUTPUT_TIMEOUT_MS} before you suspect the seam. + * on commit 20c1094. The same four cells were also run under load, with 8 spinners on 12 cores. + * The failure and the passes from that run are not worth the same. The cells ran in a fixed + * order while the load climbed from 7 to 50. The old version ran last, at the top of that climb, + * so it has a free explanation for failing and its 0 of 3 is discarded. A pass has no such free + * explanation: a cell that survives a worse condition than a fair order would have given it is + * evidence in the safe direction. So keep the three passes, each with the load it ran at: the + * fixed version 3 of 3 at load 7.42 to 18.42, the 0ms-write cell 3 of 3 at 18.42 to 23.65, the + * 1000ms-write cell 3 of 3 at 23.65 to 46.40. Above about load 20 everything here is slow for + * reasons that have nothing to do with this seam, so read the positive claim — that the read + * deadline was the whole cause — as "measured near idle on a 12-core host", and nothing + * stronger. Do not carry that raw load average to another host either: load average counts + * differently per core and per operating system, so only load per core compares. If this test + * fails on a smaller or busier machine, raise {@link #OUTPUT_TIMEOUT_MS} before you suspect the + * seam. * *
{@link #waitUntilSettled} is kept as cheap insurance against that swallow case, not because - * anyone showed it was needed. If you want to delete it, the honest test is whether you can make - * this test fail by typing early. Nobody has managed that yet. + * anyone showed it was needed. The cell that tests the swallow case head-on is the one with + * {@link #SHELL_READY_TIMEOUT_MS} at 0: input is typed at once, which is the worst case for + * "typed before the prompt is ready". It passed 3 of 3 at load 18.42 to 23.65. The wider read + * window cannot explain that pass away, because a swallowed keystroke is LOST, not late — the + * command never runs, so no amount of polling makes its output appear. So the swallow mechanism + * was tested and did not show up. If you want to delete this call, that is the cell to re-run. * *
Tagged {@code contract}; run with {@code mvn test -Pcontract}. */