[daigoba66]

> One best practice that I'd love to see spread (in our codebases too) is always naming the full feature flag directly in code, as a string (not a constant).

Can you elaborate on this? As a programmer, I would think that using something like a constant would help us find references and ensure all usage of the flag is removed when the constant is removed.

[zellyn]

One of the most common things you want to do for a feature flag or metric name is ask, "Where is this used in code?". (LaunchDarkly even has a product feature that does this, called "Code References".) I suppose one layer of indirection (into a constant) doesn't hurt too much, although it certainly makes things a little trickier.

The bigger problem is when the code constructs metric and flag names programmatically:

    prefix = "framework.client.requests.http.{status%100}s"
    recordHistogram(prefix + ".latency", latency)
    recordCount(prefix + ".count", 1)

    flagName = appName + "/loadshed-percent"

    # etc...

That kind of thing makes it very hard to find references to metrics or flags. Sometimes it's impossible, or close to impossible to remove, but it's worth trying hard.

Of course, this is just, like, my opinion, man!

[dgorton]

Agreed. Flags are a type of technical debt. Keeping them as full strings in the code encourages and facilitates cleanup.

This sort of programmatic naming is a dangerous step down a slippery slope.

[athenot]

Not OP but multiple code bases may refer to the same flag by a different constant. Having a single string that can be searched accross all repos in an organization is quite handy to find all places where it's referenced.

[joshuamorton]

especially when you have different languages with different rules, `MY_FEATURE_FLAG` and `kMyFeatureFlag` and `@MyFeatureFlag` might all be reasonable names for what is defined as `"my_feature_flag"` in the configuration.

Using just the string-recognizable name everywhere is...better.

[grork]

IME searching for the name of the flag name and getting 1 result is less helpful than 15 results that directly show point-of-use.