vdbergh

I think what Vondele wants to say is that since you reduce based on total itp, a user with many LTC's will get fewer resources (since they have higher itp).

Seems feasible. But it seems that while Linux can easily handle long command lines, cutechess-cli still crashes. So just excluding windows workers is not enough.

Ok fair enough. One needs a piece of code that estimates the length of the command line corresponding to an SPSA run. Other than that it is straightforward to implement.

To be honest I believe none of your reasons is compelling. On the other hand I also don't know why the worker (not the server!) is shipped with some pre-installed...

> I dont see how it's fine to have the packages directly in this repository and at the same time deny the addition of docker. So we are saying we...

> > If a package is distributable then distributing it is not a copyright violation. If the `idna` package does not contain a `LICENSE` file, then that is a problem...

I wonder if what you claim is correct. get_results first checks if the results are stale. If they are not, no lock is acquired. I think the case that the...

I would imagine staleness only applies to runs that are in the cache. Since rendering the main page does not use the cache, this could be a non-issue.

I find it confusing that `hist['Close']` actually refers to the adjusted close price and not to the real close price. IMHO yfinance should supply both. EDIT: I am now seeing...

I guess I messed up the updating process... I neglected to test it since I thought that it would work in the same way as before. But now I realize...