Document and conform log levels to assist customers in understanding the severity of different error codes

[markcoxsmith]

The following customer was confused about how they find out fatal errors with connections and channels:
https://app.intercom.com/a/apps/ua39m1ld/inbox/inbox/all/conversations/38483300001514

The following customer was confused by a non-issue 80008 indicating channel disconnection. It would be helpful to outline that this is not something to be unduly worried about: https://app.intercom.com/a/apps/ua39m1ld/inbox/inbox/all/conversations/38483300001724

┆Issue is synchronized with this Jira Task by Unito

[markcoxsmith]

As a result, it may be helpful for customers to have an idea of the severity of their issues so that they can easily diagnose how problematic their issue is.

[ably-sync-bot]

➤ Ivan Kavalerov commented:

This is an interesting one. When we get to this ticket I want to take a look at the ways we emit errors in different libraries - maybe there is a way to emit them in a more obvious way as well?

[mattheworiordan]

The loggers, when emitting errors, always include a log level, such as warn or error. I am not sure it's practical to couple an error code to the severity as it may be different depending on the context i.e. the error code maps to a particular type of problem, but that is not necessarily an indication of the severity. A good example is a timeout, a timeout may cause a message publish to fail, or trigger a retry. With the retry, we may want to show info or major log level, especially if it subsequently succeeds, whereas an error would be emitted if it resulted in failure to publish the message.

I did see a discussion recently (I think with @QuintinWillison) on the fact that our spec does not define log levels, and instead we leave that to the developer decide and to rely on the standard log levels available in that language. This is perhaps an area we want to tighten up in the spec too.

[ably-sync-bot]

➤ Ivan Kavalerov commented:

Jamie Newcomb do you know why this stopped syncing? Looks like it may because of some tags Matt has added on the original GH issue?

[sync-by-unito]

➤ Automation for Jira commented:

The link to the corresponding Jira issue is https://ably.atlassian.net/browse/SDK-2781