Tobias Nyholm

You cannot (or should not) access your database at the time when you build your container. That means you have to redeploy every time you get a new customer. What...

That is a good question. We should document that answer in the readme.

What version are you running? You should not be required to configure the guzzle client like that.

Could it be related to: https://docs.github.com/en/actions/reference/events-that-trigger-workflows#triggering-new-workflows-using-a-personal-access-token (From [stefanzweifel/git-auto-commit-action docs](https://github.com/stefanzweifel/git-auto-commit-action#commits-of-this-action-do-not-trigger-new-workflow-runs)) If so, I think it is a bug.

I’ll be happy to.

Hey. You are correct. They were removed earlier than the deprecation message said. However, it is not a SemVer violation. SemVer acts differently before 1.0.

When testing this in bash, then `foo` will be moved inside `bar`. So Im not sure that I would expect `bar` to be overwritten as you say.

Adding methods to an interface is a BC break. I cannot merge this.

Btw. Version 0.10.0 will support AsyncAws and that switches to multipart uploads automatically. It will not have a memory foot print that is related to the part size.

It did help me. I also managed to squeeze in an extra header that I didn’t show in my workaround example.