Feature request: main implementation, randomization, middy middleware, correlation ID's, optional: object merging

[saragerion]

Description of the feature request

Problem statement

There are some areas that should be revisited or implemented.

Summary of the feature

1. We should update the implementation of randomisation of sampling - currently implemented like this:

  private setLogsSampled(): void {
    const sampleRateValue = this.getSampleRateValue();
    // TODO: revisit Math.random() as it's not a real randomization
    this.logsSampled = sampleRateValue !== undefined && (sampleRateValue === 1 || Math.random() < sampleRateValue);
  }

Not optimal.

2. Implementation of middy-compatible middleware logic to match the one of the decorator which is already existing.

3. Correlation ID's propagation - see Feature request: correlation ID's propagation #129

4. Optional: replace lodash library with in-house custom code

Code examples

Benefits for you and the wider AWS community

Describe alternatives you've considered

Additional context

Related issues, RFCs

#129

[bahrmichael]

Writing down my understanding from our discussion yesterday:

1. We have tests where Math.random yields a spread (or entropy) that's too large (i.e. 50 to 150). We should find some random that reduces the entropy, preferably to something like 90 to 110. This should not increase the bundle size significantly.
2. Decorators only work for classes, but many people use functional programming with TypeScript, and as far as I understand we can't decorate functions (which are not part of classes). Therefore we want to investigate if we can add logging/tracing/metrics via middy middleware. Example pseudo code below:

app
    .use(cors())
    .use(tracing()
    .use(metrics({coldStarts: true})

3. No further details, as there's a full issue: Feature request: correlation ID's propagation #129
4. Lodash was used because it was easy to get started. It's a big library though, so we'd like to replace it with something simpler and smaller.

[dreamorosi]

Regarding number 4, a quick win until we have our own implementation could be to switch to lodash.merge and lodash.clonedeep instead of importing the full package.

Just for context, these two are the functions that I saw are used in the Logger and that are exported as separate packages.

[saragerion]

@bahrmichael your understanding is 100% correct.

@dreamorosi well spotted! I wasn't aware that separated packages formats exist.

[github-actions]

⚠️ COMMENT VISIBILITY WARNING ⚠️

Comments on closed issues are hard for our team to see.
If you need more assistance, please either tag a team member or open a new issue that references this one.
If you wish to keep having a conversation with other community members under this issue feel free to do so.