tore g. eschliman

to their credit, in local they actually did uncover issues (I didn't understand USR1's purpose at all really until I dug in and noticed it wasn't defined behavior), but they...

@wlipa @PikachuEXE 's explanation [here](https://github.com/puma/puma/pull/3643#issuecomment-2785024761) is correct, which I tried to explain in the documentation [here](https://github.com/puma/puma/pull/3643/files#diff-39997921b37af5764c26e470d9d98f64907f6d7dd74bcd03a1d333551fb6d81bR53); please let me know if the explanation is too dense or poorly worded and...

AFAIK puma doesn't track a request count per cluster anywhere, they are all per-worker; regardless I have updated the doc in my branch of final changes, just waiting for any...

that is correct, yes; the only other nuance is that when you refork from a mold, all the new workers inherit the mold's request count, so after the first refork...

I am running a branch at https://github.com/toregeschliman/puma/pull/13/files for any final feedback, IMO it's easier to review when you don't have a moving target and hopefully there's nothing substantive left.

anyone have a chance to dogfood and got feedback they could share?

@PikachuEXE how many workers are you running? there should be no additional cost to extra refork cycles (new molds replace old ones so they don't accrue over time) but the...

@MSP-Greg sorry for the nudge, but anything I can do to move this forward? It's getting harder to justify running off a personal branch for as long as we have....

The alternative I'd hope for is the feature gets merged so we could run off of master and then eventually a normal versioned release; it makes normal maintenance easier and...