Matt Traudt authored Jun 25, 2018 and juga committed Jul 05, 2018

User should configure logging via the [logging] section
in their config.ini.

User can change the log level using --log-level, or in the config
with level, to_file_level, and to_stdout_level.

User can change the log format in the config with format,
to_file_format, and to_stdout_format

An example debugging format is included as a comment in config.default.ini

config.log.default.ini is majorly reorganized. For the most part, if
something doesn't need to be specified in this file, it has been removed.
For example, we don't have to specify any args to handler_to_file. The
beginnings of syslog configuration stuff remain as comments.

As part of configure_logging, force logging to stdout if we don't think
we can log to file.

Actually set some parameters to RotatingFileHandler so it rotates when
files reach 10 MiB and keeps 100 backups.

Each sbws command (like scanner and generate) gets its own log file. I
did this because I didn't want 'sbws scanner' to be running 24/7 and an
'sbws generate' call to log to the same file. That would get confusing.

but not allowing /dev/stdout

to perform usability tests. This will avoid raising IndexError
when choosing an exit.

GH: closes #218

Since version 3.4.2 gives error on install_layout and it is not
the default in stable OS distributions

**Important changes**:

- Result format changed, causing a version bump to 4. Updating sbws to 0.5.0
  will cause it to ignore results with version less than 4.

Summary of changes:

- Keep previously-generated v3bw files
- Allow a relay to limit its weight based on
  RelayBandwidthRate/MaxAdvertisedBandwidth
- 1 CPU usage optimization
- 1 memory usage optimization

Check that there are results before trying to get the results
for an specific relay. This would avoid raising TypeError
when there are not results.

log the error, ingore the line and continue

Matt Traudt authored Jun 22, 2018 and juga committed Jun 24, 2018

The chance of us needing a refresh at this point is TINY, but it's probably
better to return a list of relays stale by a millisecond than it is to cause
the user's program to crash.