Paul Moore

> If we could just export MAX_PIP_RESOLUTION_ROUNDS=50000 for example (and that would override now hard coded try_to_avoid_resolution_too_deep = 2000000 definition) that would give us control over it. Please read some...

> I think the only thing needed would be the ability to control that try_to_avoid_resolution_too_deep = 2000000 externally. Please read the other issues. This has been extensively discussed, and the...

> I'm not opposed to the idea that we should reduce that number. :) To be clear, nor am I. I am uncomfortable about making it user-settable, though, as we'll...

> Pip doesn't give enough useful information in the logs to figure out which packages are the causing the pathological backtracking. I think there's a related issue here as well....

> I wonder if it’d make sense to do optional telemetry only to people who set a custom resolution round. Do you mean actual telemetry here (as in, pip uploads...

> Instead we would like to know why pip was not installing the latest version of the requested package, as we would have expected it to. And at that point...

> I'm personally fine if we implemented a `--no-backtracking-in-resolve` flag or something. Agreed, that seems like a relatively easy option to describe/understand. I have my reservations about whether in practice,...

> What problems would come from reading the value for try_to_avoid_resolution_too_deep from an environment variable (if exists and use default if not)? What am I missing? Feel free to experiment...

> So, I don't mean to nag, but what is the blocker / problem with opening an environment variable into that value? Nobody willing to do the work to get...

No that is *not* the "official pypa/pip recommendation", and frankly you're coming across a bit passive aggressive here. I was responding to @kariharju and in particular to their "what is...