Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Design and implement pluggable exposure logging #23

Open
eric239 opened this issue May 28, 2015 · 7 comments
Open

Design and implement pluggable exposure logging #23

eric239 opened this issue May 28, 2015 · 7 comments

Comments

@eric239
Copy link
Contributor

eric239 commented May 28, 2015

While different consumers of p4j will likely have different logging implementations, it would be nice to have a pluggable framework that abstracts the "what" to log from "how" it gets logged

@eric239
Copy link
Contributor Author

eric239 commented Jul 27, 2015

Adding a bit more specifics.

First, let's define an interface for logging implementation:

public interface Planout4jLog {
  void log(Planout4jLogRecord r);
}

Default implementation will utilize SLF4J.

The log record class will encapsulate exposure information per namespace, including (but not limited to)

  • namespace name and salt
  • inputs
  • experiment name and salt
  • boolean indicating default experiment or not
  • experiment's planout code and its checksum
  • all parameter names, salts, and evaluated values

Let's also define a separate interface for serializing an instance of log record:

public interace Planout4jLogRecordSerializer {
  String serialize(LogRecord r);
}

Default implementation will produce JSON.

Each YAML namespace config will be extended to include the following logging-related properties:

  • FQN of log impl class
  • FQN of serializer class
  • logging on/off (boolean)
  • log default experiment exposures (boolean)

@reikje
Copy link

reikje commented Jul 30, 2015

Looks good I think. At which point are you planning to create the Planout4jLogRecord?

Example based on the simple config file. A user can either see the default button #000000 | Register or buttons that are permutations of #ff0000, #00ff00, Join now!, Sign up.

A client code example might look like:

NamespaceFactory nsFact = new SimpleNamespaceFactory(); // done once on startup?
Namespace ns = nsFact.getNamespace("test-ns", Collections.singletonMap("userid", 123).get(); // done for each user
String buttonText = ns.getParam("button_text", "default"); // can we populate a full Planout4jLogRecord here?
String buttonColor = ns.getParam("button_color", "default"); // or are the params evaluated lazy?

Does any of the calls above initiate a callback to the log method or will it be an extra (explicit) call?

@eric239
Copy link
Contributor Author

eric239 commented May 20, 2016

In the above example, the 2nd line will trigger logging.
There's an implicit assumption that application will "cache" Namespace instance (which in turn has all the parameters computed and cached after creation, so any of getParam calls will not trigger logging)

BTW, the implementation of the above suggested framework is about to begin 😄

@ivandd
Copy link

ivandd commented May 24, 2016

@eric239 it's great that you have found time to do this. Looks like the log record has what one would normally want to log.

eric239 added a commit that referenced this issue Jun 1, 2016
eric239 pushed a commit that referenced this issue Jun 1, 2016
#23 pluggable logging with basic impl
@nkapur
Copy link

nkapur commented Jul 26, 2017

Quick question on this: I see 1.2 was released in May 2016 - so I'm assuming these changes did not make it in yet. Is there a plan to release a 1.3?

@eric239
Copy link
Contributor Author

eric239 commented Jul 26, 2017

@nkapur this project would benefit from a new maintainer(s). I (the original maintainer) have moved on and don't use planout4j in any of my current work.

@ryanaylward is there someone at Glassdoor interested in "adopting" planout4j?

@nkapur
Copy link

nkapur commented Jul 26, 2017

I think having configurable Logging would be a good stopping point. I believe active maintenance might not be required medium or long term.

We at GoFundMe are building components for venture projects without having to rely on the internal planout micro-service.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

4 participants