For some reason printing the Result made the error string print twice,
along with some annoying newlines. Extracting the error first gets the
expected result (just one explanation, no newlines)
Just a few more places where we can drop the .unwrap() call in favor of
`?`.
Also include a fix to the log file handling: don't open the file twice.
Writing to two fds would result in one message overwriting another.
Presumably `File.try_clone()` reduces down to `dup` on Linux.
This is a first attempt at a new error-handling strategy:
- Use anyhow::Error as the first choice for easy error handling
- Use thiserror to generate local error types for anything that
needs it (no error type is available to us) or will be inspected
or matched on by higher layers.
Resolve some basic warnings from clippy:
- useless conversion to the same type
- redundant field names in struct initialization
- redundant single-component path imports
We don't need the nightly compiler, but there's no reason it shouldn't
compile without warnings, either. I don't know why stable doesn't warn
about these, but it's cheap to suppress them.
Using the hash should allow us to change the remote repo and propagate
that change to user builds without that change becoming visible at a
random time.
It's unfortunate that we can't declare this dependency once in the
top-level Cargo.toml; that feature request is rust-lang rfc 2906.
When postgres sends us a keepalive message, send a reply so it doesn't
time out and close the connection.
The LSN values sent may need to change in the future. Currently we send:
write_lsn <= page_cache last_valid_lsn
flush_lsn <= page_cache last_valid_lsn
apply_lsn <= 0
