ppigazzini

> Yes. But 2. does not show in your runs. However it is reported in the event log. > > It think sometimes the parsing of the "Score" lines fails....

> edit: the max length is 32767 characters, https://devblogs.microsoft.com/oldnewthing/20031210-00/?p=41553 The command line is over 235k characters

> @ppigazzini Check the cutechess-cli v1.3.0 output, no lines with "Score", and cutechess-cli segfault as well. > > Yes but in your runs the worker seems to handle it gracefully....

I ran few times in WSL2 that long command line replacing `cutechess-cli` with `fast-chess` and the tournaments always finished, I given to Disservin (`fast-chess`'s author) the log of one tournament....

@maximmasiutin `pyshbench` uses `stockfish bench`, shortcut to `stockfish bench 16 1 13 default depth` Using depth 20 with `stockfish bench 16 1 20 default depth` gives more stable nps (in...

Consider if the code could be clearer using `Enum` or `StrEnum` (I don't know if it's compatible with `vtjson`) https://docs.python.org/3/library/enum.html note: PROD is running with python 3.12.3

> This PR is almost totally untested because > > * testing the purging code is tricky; > * I lack the resources to generate tasks with meaningful residual @peregrineshahin...

DEV running with the PR. The finished tests have the same residuals on PROD and DEV (checked the old ones as well). There are some on-going tests with time odds...

Internal server while purging https://dfts-0.pigazzini.it/tests/view/6783db7ff90a2cc8d20e5a45 Click to view ```yaml Jan 12 17:59:37 dfts-0 pserve[90464]: The run object 67823a24a31c4c13e83518a8 does not validate: run['bad_tasks'] is missing (this is likely not an error...

I created that run after merging the PR, so it should use the new schema. Enjoy your dinner ;)