Releases: karl-gustav/runlogger
Releases · karl-gustav/runlogger
Release version
Removed the *j methods in favor of just using log.Field() everywhere Now all the remaining logging methods can take log.Field() anywhere (the *f methods need to have the formatted string first) Never liked the *j methods in the first place so this is a huge win.
<severety>j loggers now take zero to many log.Field() methods
Making a non-backwards compatible change
v0.1.1: Add ServiceContext to structured log output
Because that way we can group the Errors from a service in the google cloud Error Reporting tool
v0.1.0: Add @type the the different error severeties
Because we want the errors to show up in the Google Cloud Error Reporting tool. This achieves that. We could also add `stack_trace` at the root of the logEntry but that is not always desired. Plus we have the source location.
fmt.Sprint --> fmt.Sprintln
Change standard log functions from fmt.Sprint --> fmt.Sprintln Because Sprint doesn't add a space between the arguments when it prints
v0.0.9: Remove relative path from filenames
Beacuse it is much easier to read [main.go:33] than [/home/kgr/git/test_runlogger/main.go:33]. And it makes even less sence when using this in the cloud...
v0.0.7: Fix: still not logging
Probably buffering again...
v0.0.6: Simplify: Only have one method to write logs
Because too much duplicate logic to have both of them when they are so similar. Also fixed a bug with a missing flush of stdout
v0.0.5: Logging an error instead of truncating the message
Because the other logic didn't have a good way of checking the size of the object you where potentially trying to serialize into the log.
v0.0.4
Setting hard limit on message/TextPayload to 100000B Because GCP has a hard limit on 100KB (102400B) and I use the remaining 2400B on the other fields like timestamp and so on.