Richard Levitte

Rebased on a fresh master. Now that the other config system changes are in, this one looks much easier. The Travis change is only a temporary commit for testing purposes,...

Yes. Question is if it's relevant for 1.1.1 or not. It certainly would make certain things easier, but it's not strictly necessary yet. @dot-asm, your thoughts?

Actually, this is quite a fundamental change to how the config database is treated. Essentially, it implements lazy evaluation and recursion in a way we don't have right now. So...

There are parts of this PR that aren't relevant any more, as @dot-asm implies. I was thinking of the OpenSSL::ConfigHash thing, mostly. That's a fairly central part

Yup. Please keep this PR live for the moment, though. It helps me remember...

If others could actually spend some time reviewing the STORE stuff I do, that might take it a step forward. One of the PRs I've worked on lately (mostly because...

> AFAIK you must use ossl_intmax_t instead of intmax_t to avoid breaking some platforms. That's beside the point for the macos build failures, the errors are all about `int64_t` and...

I think there should be. The effects that I desire are: - that this is clearly mentioned in the `-help` output. - that any use of such an option generates...

I think that most of all, we need a common *method* for deprecating options.

I didn't see the earlier notification. I think we need to consider this again.