Richard Schneeman

A workaround is to have it return `InvalidMetadataAction::DeleteLayer` after the `todo!()`. I'm not sure if we can otherwise improve the experience. Can we implement IntoAction for `()` ?

I've got a "yes and" for this issue. My experience with debugging builds on Heroku is that usually you only know you need the verbose output AFTER the build has...

Thanks for the fast feedback! > We do some things like walking a directory to log the contents of what's in the buildpack and some other folders Excellent point. We...

I'm leaning towards not doing this. More: Looks like it's more of a convention: https://superuser.com/questions/944958/are-http-proxy-https-proxy-and-no-proxy-environment-variables-standard. I happen to have a PR today to update this function (but not with your...

That was fast. In another Slack thread, I noted that value never seemed to be false at that location in the server.rb. I thought it was a case where we...

Good news, bad news. It looks like that logic works correctly now, but we're still seeing one request with a really long tail. With 5 puma threads, and default max_fast_inline...

I was thinking about this in the ride home after dropping my kids off at school. I think I know where that 2-second request is coming from. I like the...

I forgot all about this PR, and funny enough we landed something very close to it in https://github.com/puma/puma/pull/3794. That PR landed on a slightly different heuristic, it uses "fast inline"...

I was bit by this behavior recently as well. I had it on my backlog to file an issue about it, but I see Ed already has. Thanks Ed. I'll...

Short: I agree "anyhow" isn't great. Something implementation hinting like `include_error_source` only makes sense in half the internals. I like a "behavior" focused feature name. I'm leaning towards `custom_caused_by` to...