Compilation with rocksdb 3, make rocksdb backend default and optional

[eile]

@lasch did you see this?

[lasch]

yes. I did. Sorry, delayed by holidays here. I'll check ASAP.

[eile]

Ah true. I forgot the turkey fest. No hurry.

[lasch]

Regarding the FxLogger changes. I'm not yet sure about the consequences of removing the PK_LINUX macro and rely on just linux. When building for CNK, we might still have linux defined and might not want that in some places. So I'll partially merged this PR for now (leaving f4a9e95 open for now)

[eile]

Ok, but non-selfcontained defines in headers are bad, because downstream projects need to implement logic they have no awareness of. Can we set PK_LINUX automagically?

[lasch]

I think, automagical setting should be the goal. yes. It's possible that it's already contained, however I want to double check first.

[eile]

This is somewhat important now, since it trips our unit tests and CI.

[lasch]

I've merged the linux-macro code.

[lasch]

Pulling the latest 2 commits, I'm seeing 2 commits with the same comment now. What's the story about that?

| * 5b4f6fc - Compilation with rocksdb 3, make rocksdb backend default and optional (5 days ago)

• | f7535f5 - Compilation with rocksdb 3, make rocksdb backend default and optional (7 days ago)

[eile]

I only see the last in this PR. try to hard-reset to origin/master and repull?

[eile]

Ah, you cherry-picked it already. Let me rebase and update this PR.

[eile]

Done.

[eile]

Cherry-picks change the SHA and make it a different commit.

[lasch]

I didn't cherry-pick it. I forward-merged the initial pull request to commit 5b4f6fc. Leaving only f4a9e95 unmerged (see comment 5 above). Thanks for cleaning up.