Handling monads {:ok, result} and {:error, description} can be very tricky in a complex systems.
Moreover, it is very difficult to track the history of the error and it's influence. Unfortunately there are no built-in tools for that.
Forensic is pass-through error protocol and a set of macros, aimed to sctructurally register the description of the error in the context, where it had occured and collect all the collateral errors.
Forensic's protocol fully follows {:error, description} ideology.
Forensic.error(error, description \\ :no_description, contents \\ nil) will transparently incapsulate all {:error, description} or even any arbitary error given.
def deps do
[
{:forensic, "~> 0.2.2"}
]
endForensic.error(...) returns the error tuple in the format:
{:error,
%Forensic.Errors{
history: [
%Forensic.Error{
bindings: [],
contents: nil,
description: :no_description,
function: "test Forensic.Errors protocol detection/1",
module: ForensicTest
},
%Forensic.Error{
bindings: [],
contents: nil,
description: :no_description,
function: "test Forensic.Errors protocol detection/1",
module: ForensicTest
}
]
}}Basic scenario
import Forensic
...
Forensic.error(error_to_describe)
...The full-on scenario is:
import Forensic
def Foo(some_var) do
with {:ok, result} <- Foo2(some_var) do
{:ok, result}
else
Forensic.last?(:error_signature) -> do_something_for_this_error_signature()
Forensic.last?(:error_signature_2) = error -> do_something_for_this_error_signature2()
# Register unrecoverable error afterwards
Forensic.error(error, :error_description2)
error -> Forensic.error(error, :error_description)
end
endForensic.last?(error) extracts the last error description from like was in {:error, description} providing a way to handle different errors by their signatures as well.
MIT