A simple logger tree-based filtering #204
Conversation
|
Very useful when using the SLF4j bridge (#202) |
| def filterByTree[A](root: LogFilterNode): (LogContext, => A) => Boolean = | ||
| (ctx, _) => { | ||
| val loggerName = ctx.get(LogAnnotation.Name).flatMap(_.split('.')) | ||
| val logLevel = findMostSpecificLogLevel(loggerName, root) |
There was a problem hiding this comment.
did you check what will be performance impact of this?
There was a problem hiding this comment.
Not yet, I can add some benchmarks to have an understanding. And then it could be probably improved with a cache or something.
| } | ||
|
|
||
| private def buildLogFilterTree(rootLevel: LogLevel, mappings: Seq[(String, LogLevel)]): LogFilterNode = { | ||
| def add(tree: LogFilterNode, names: List[String], level: LogLevel): LogFilterNode = |
There was a problem hiding this comment.
we probably should use Chunk instead.
There was a problem hiding this comment.
Yeah, but note that the LogAnnotation.Name has a type of List[String] too
There was a problem hiding this comment.
yeah, I forgot about. we should probably change that in the future
|
I created some benchmarks and also wrote a simple For me this looks like the tree-based filter is comparable to a hand-written one (for a few nodes in the tree, which is the normal use case to my experience - thinking of the number of differently configured loggers in a typical log4j config for example). |
| import scala.annotation.tailrec | ||
|
|
||
| object LogFiltering { | ||
| case class LogFilterNode(logLevel: LogLevel, children: Map[String, LogFilterNode]) |
|
@jdegoes I believe this is ready to merge |
Introduces a simple hierarchic name-based filtering method that is similar to other logging systems. The filter is defined by a list of names tied to minimum log levels.