Tudor Brindus

Superseded by https://github.com/DMOJ/online-judge/pull/2028.

Closing this issue due to inactivity, if the above message didn't resolve things feel free to reopen it.

Superseded by https://github.com/DMOJ/online-judge/pull/2021.

Closing due to inactivity, if the issue persists despite addressing the above comment, feel free to reopen this.

Closing because while nice, this change makes /submissions dramatically slower when the submissions table doesn't fit in RAM :disappointed:

Thanks for reporting. Could you try: * running with `MAGIC_TRACE_DEBUG=1 magic-trace ... -z-print-perf` * run the `perf` commands yourself (you might want to use `attach` with a longer-running process) *...

Thanks for trying that. What's your `perf` version? Snapshot-at-exit (which is necessary to get traces of very short-lived processes) was technically introduced in `perf` 5.4, but we've seen it be...

Ref `-immediate-stop` from https://github.com/janestreet/magic-trace/pull/79, which used `--filter` internally.

If we decrease the timing resolution in software, we can also request lower timing resolution from the hardware, for lower overhead. Refs: * https://github.com/janestreet/magic-trace/blob/e9d5d2683da18ed317a28b3f9531a249c1a57443/src/perf_tool_backend.ml#L52-L61 * http://halobates.de/blog/p/432

Fuchsia supports inserting binary blobs to the trace file: https://fuchsia.dev/fuchsia-src/reference/tracing/trace-format#blob-record We could include the `perf.data` and associated files there (as a tarball?), I think, to allow later processing.