Is there a reason not to call Exception.blame automatically?

[axelson]

Is there a specific reason that Exception.blame/3 is not called automatically by Honeybadger? It does say that "This operation is potentially expensive, as it reads data from the file system, parses beam files, evaluates code and so on." but I think this should at least be optional when using honeybadger since it provides a much richer error.

It can convert an error like this:

no function clause matching in Messages.handle_message_created/2

into:

no function clause matching in Messages.build_notifications/3

The following arguments were given to Messages.build_notifications/3:

    # 1
    %Chat.Message{__meta__: #Ecto.Schema.Metadata<:loaded, "messages">, ...snip large struct value}

    # 2
    nil

    # 3
    nil

Attempted function clauses (showing 3 out of 3):

    def build_notifications(message, thread, -%Messages.MessageContainer.MessageContainerImpl{container_id: post_id}-)
    def build_notifications(%{id: message_id}, _thread, -%Communities.Post{message_id: message_id}-)
    def build_notifications(message, thread, -%Communities.Post{} = post-)

Might be related to #123

[sorentwo]

@axelson There isn't a good reason at this point. Historically the primary reason is that we don't require a kind argument for notify/3 and we treat all exceptions like an error when normalizing.

I think it would be a good change to get in though. I can look into it next week, or if you're interested a PR is certainly welcome =)

[axelson]

This would be really useful for us so I've created a PR #219