bmc / grizzled-slf4j

Simple Scala-friendly front-end to SLF4J

bmc

Just some quick details:

• I'm a software guy currently employed by Databricks.
• I have my own (currently mostly inactive) company, ArdenTex
• My personal web site, such as it is, is clapper.org/bmc
• Find me on Mastodon at https://jawns.club/@bmc

Other links are in my GitHub profile.

grizzled-slf4j's Issues

apply[T](implicit manifest) to Logger object

I would suggest to added a new apply method to Logger object like:

def apply[T](implicit m: Manifest[T]): Logger = apply(m.erasure)

So one would use:

val logger = Logger[MyClass]

how to configure grizzled to log different log levels to different log files?

I want to use two different log levels, trace and info, and log them to different log files. I saw that this is possible in Java, following this StackOverflow question.
Is there a way to do so in Grizzled with an SBT project? At the moment I can only output one log level to one file by creating a file "simplelogger.properties" following this article.
Any help would be appreciated.
Thank you

Could you publish to central repository?

I've found grizzled-slf4j 1.0.2 at Bintray only. Therefore we have to add the following URL to resolvers.
http://dl.bintray.com/bmc/maven/

Could you publish to the central repository?

Inline optimization warning

I'm receiving the following warning when compiling with "scalac -optimise". I don't think its due to my usage, though.

This is with v0.6.6 and Scala 2.9.1

[WARNING] Foo.scala:101: warning: Could not inline required method warn because it is unsafe (target may reference private fields).
[WARNING] log.warn("Fake log message ")

Remove `()` in `Logger.apply[C: ClassTag]()` method?

The Scala 2.13.3 deprecates auto-application of (non-Java-defined) methods with a single empty parameter list: scala/scala#8833.

So the below example in the documentation doesn't work anymore (without suppressing the warning):

class Foo {
  val logger = Logger[this.type] // or Logger[Foo]
}

Would it better to remove the () in Logger.apply[C: ClassTag]() method here:
https://github.com/bmc/grizzled-slf4j/blob/master/src/main/scala/grizzled/slf4j/slf4j.scala#L434 ?
Then we could still use Logger[this.type] in our code.

Documentation error: sbt setup

On the frontpage, http://software.clapper.org/grizzled-slf4j/, it says that for scala 2.9.x one should use:
libraryDependencies += "org.clapper" %% "grizzled-slf4j" % "1.0.1

this does not work, I believe the correct version is:
libraryDependencies += "org.clapper" %% "grizzled-slf4j" % "0.6.10"

do you think an api like trace[A](x: A): A so that i can return a value while tracing it as well is proper for this lib?

Does it support Markers?

Using logger in a Serializable class

Is there a recommended way to use a grizzled.slf4j.Logger in a Serializable class?

I'm currently mixing in Logging to my class and getting serialization errors due to the non-serializable _logger member:

[info]   Cause: java.io.NotSerializableException: grizzled.slf4j.Logger
[info] Serialization stack:
[info] 	- object not serializable (class: grizzled.slf4j.Logger, value: grizzled.slf4j.Logger@19c58b6)
[info] 	- field (class: org.my_group.my_class, name: grizzled$slf4j$Logging$$_logger, type: class grizzled.slf4j.Logger)
[info] 	- object (class org.my_group.my_class, DepthMap(MapPartitionsRDD[27] at map at DepthMap.scala:149))
…

This is in the context of a Spark application that is serializing a task closure, fwiw.

This SO has general discussion about this kind of issue but nothing that will specifically unstick me here, afaict.

Split into API and implementation

Split the API into two parts:

• A set of API traits, which are logging framework-agnostic.
• An SLF4J implementation of those traits.

Allows alternate implementations (e.g., on CLR) that aren't tied to SLF4J.

version 1.0.1 depends on scala-library 2.10.0-RC1 in compile scope

I'm a Scalatra user and Scalatra depends on this library. The latest stable version 1.0.1 seems to depend on scala-library 2.10.0-RC1 in compile scope.

http://search.maven.org/remotecontent?filepath=org/clapper/grizzled-slf4j_2.10/1.0.1/grizzled-slf4j_2.10-1.0.1.pom

<dependency>
<groupId>org.scala-lang</groupId>
<artifactId>scala-library</artifactId>
<version>2.10.0-RC1</version>
<scope>compile</scope>
</dependency>

Actually it's not a big problem, but ideally this dependency is not needed. Do you have any plans to release new version?

make slf4j-api "optional"

Can you make the slf4j-api dependence in build.sbt "optional" rather than none/?

Otherwise, things like the sbt web plugin are going to stuff the library into wars they create and that's going to play havoc with servlet container class loaders which will find it both at the webapp and server level and not be happy.

Unless, of course, there's a better way to fix this ...

is there warn(somOperation) which returns the someOperation result in case its None?

for example I have this
for {
user <- UserRepository.findById(1) // do you ever need to log warning here if None? what is your best practice for it here?
gender <- user.gender
} yield gender // results in Some("male")

I thought of using something like

user <- warnIfNeeded(UserRepository.findById(1))

now instead of writing my own warnIfNeeded would be great to have it in a shared lib... what do you think?

Support Logging trait

See, for instance, this search.

Thanks to Sami Dalouche for the suggestion.

Logging fails with Scala 2.10.0 Final

When used with 2.10.0 all logging calls fail to instantiate a logger with a java.lang.NoClassDefFoundError: scala/reflect/ClassManifest thrown from the call to getClass in the logger initialisation.