Question: Why do code examples not follow lower_case variable names and UPPER_CASE constants?

[couling]

Apologies for asking about a really boring topic.

I noticed in the README, metrics stored globally with upper case naming:

client_python/README.md

Line 19 in d6e08e3

REQUEST_TIME = Summary('request_processing_seconds', 'Time spent processing request')

This is a bit confusing for many readers because across multiple languages (including python) UPPER_CASE names are use to denote a constant (a value that does not change) and lower_case names are used for variables (that do change).

Likewise in the code base there are things like REGISTRY that are uppercase for non-constant items:

client_python/prometheus_client/registry.py

Line 168 in d6e08e3

REGISTRY = CollectorRegistry(auto_describe=True)

Is there a specific reason that these examples a reusing upper case names even though it conflicts with common standard and doesn't match the metric name?

I'm just asking because if there is a good reason, this might actually be worth documenting.

Apologies again for the trollish sounding question.

[csmarchbanks]

These decisions were made before I worked on this project, but I believe it is because the metric object should not change, e.g. you should not replace REQUEST_TIME with a new Summary. Similarly the default registry should not be replaced with another registry. You can interact with it, but should not overwrite it.

There could be a different reason, but that is my thought.

[couling]

Hmmm. It may have been the reason, but it's a misunderstanding of "constant". For example a list object that is never replaced, but continually modified appending and clearing is not constant.