- Initial "MVP" version
- Added PyYAML as a package dependency
-
Project renamed to "Sherlog"
-
Exception type in the monitor CLI is not renders in upper case
-
Exception type and exception repr are now separated text fields in the DB
-
Stacktrace is not a separated text field in the DB
-
Thead ID and Process ID are not field of type BIGINT in DB Line is now of type INTEGER in DB
-
Some DB field names is shorter now, type renamed to lvl (message level), changed some data types
id SERIAL NOT NULL PRIMARY KEY app TEXT NOT NULL ts TIMESTAMP NOT NULL ts_rel INTEGER DEFAULT NULL lvl TEXT NOT NULL message TEXT DEFAULT NULL module TEXT DEFAULT NULL function TEXT DEFAULT NULL line INTEGER DEFAULT NULL path TEXT DEFAULT NULL file TEXT DEFAULT NULL pid BIGINT DEFAULT NULL pname TEXT DEFAULT NULL tid BIGINT DEFAULT NULL tname TEXT DEFAULT NULL logger TEXT DEFAULT NULL ex_type TEXT DEFAULT NULL ex_repr TEXT DEFAULT NULL stack TEXT DEFAULT NULL
-
filter -> types from monitor config renamed to filter -> levels
-
Some minor changes to monitor output styling for better readability ('\n' after exceptions, etc)
- Sherlog is now compatible with Python 2.7 (therefore type annotations in source code removed)
- Few fixes on Python 2.7 support (unicode issue, stack_info key error)
- Fixed critical bug in worker
- Fixed issue with log messages that contains formatting symbols such as '%' or {}
- format_style kwarg was removed from the sherlog.set_logger() signature
- Sherlog now creates functions in DB within the defined schema and not in public schema
- Added new parameter to client config main section. If 'dummy' is true, then no Redis handler will be attached to the logger. It may be a good idea during development, when you don't want to spam logs to Redis instance and DB while debug locally but still want to keep your code base as if Sherlog is there.
- Project dependecies versions are not strictly fixed anymore