Allow changing log level after connection is established. Because
log level and loggers can be set independently, it is now possible
to have a log level above none when there is a nil logger. This
means all log statements need to check for nil logger and an
appropriate log level. This check has been factored out into
*Conn.shouldLog.
Allow replacing logger after connection is established. Also
refactor internals of logging such that there is a log method that
adds the pid to all log calls instead of making a new logger object.
The reason for this is so pid will be logged regardless of whether
loggers are replaced and restored.
rxMsg() has a guard clause that checks if the msgReader has already
encountered an error. It should return that pre-existing error. It
was incorrectly returning a local `err` variable instead of `r.err`.
There's no need to read 1 byte and then immediately read 4 more, rather
than just reading 5 bytes to begin with. Also, with this change rxMsg is
no longer swallowing an error from ReadByte.
Benchmarks revealed that it is no longer performant enough to pull
its own wait. Using go_db_bench to copy JSON results to HTTP responses
it was ~20% *slower* for ~4BK responses and less than 10% faster for
+1MB responses.
The the performance problem was in io.CopyN / io.Copy. io.Copy
allocates a 32KB buffer if it doesn't have io.WriterTo or io.ReaderFrom
available. This extra alloc on every request was more expensive than
just reading the result into a string and writing it out to the response
body.
Tests indicated that if MsgReader implemented a custom Copy that used a
shared buffer it might have a few percent performance advantage. But the
additional complexity is not worth the performance gain.
Jack Christensen
Blake Gentry