Cole Miller

Closing because the execinfo.h API doesn't read DWARF info, hence you can't get a helpful backtrace unless you postprocess it with addr2line, which makes this feature less convenient for quick...

I've been looking into how to implement this. My initial plan was to have dqlite spawn a dedicated thread on startup, communicating with the main thread via a queue, and...

> but first I need to understand to what extent SQLite sychronizes its own calls to VFS functions -- i.e., imagining for the moment that we never touch the VFS...

> The VFS interface basically maps more or less directly to the equivalent syscalls provided by the OS, for instance xRead() maps to read(), and virtually has the same expectations...

Yeah, I looked into using VACUUM to reset/empty a database in #435 and concluded that a prerequisite was proper support for attaching additional databases to a dqlite-managed connection, see https://github.com/canonical/dqlite/pull/435#issuecomment-1331257989....

VACUUM needs to be coordinated using a Raft log entry, since it affects page numbers in the future.

Would this just be a cleanup, or does it unlock some possibilities that the current message format prevents?

> In general I think we should always only add fields, that makes everything easier, so the current approach should be enough. Yeah, this is how I feel. I think...

Ah, I see, because we want the support to be in place for a while before making use of it, so that we're justified in assuming that all servers in...

Okay, going to experiment with automated reformatting/rewriting now...