Without statement for Results

[markspanbroek]

Supports without statements that provide access to the error in a Result.

Syntax:

  without value =? r, error:
    # use `error` to get the error from r
  # use value

[markspanbroek]

An alternative syntax that was considered but not implemented:

  without (value, error) =? r:
    # use `error` to get the error from r
  # use value    

The reason for not going with that syntax is that it doesn't combine bindings very well:

  without ((value1, error1) =? r1) and ((value2, error2) =? r2):
    # is error1 set here?
    # is error2 set here?

[dryajov]

Nice! I like the syntax.

Is it possible to leave out the , error variable and have it behave same as the previous versions for backwards compatibility?

[emizzle]

LGTM, great job! Keen to know the answer to the above question as well, as I think it would be a good feature for backward compatibility.

[markspanbroek]

No worries, this is a backwards compatible change. The version of without with two parameters (expression and body) is still there, and I intend to keep it because it's what you use when you're dealing with Options instead of Results.

[arnetheduck]

fwiw, a similar construct exists in the form of template valueOr, in Result itself - as part of the general refresh happening there, there's a corresponding template errorOr as well now - also, lmk if there are any primitives needed in Result itself that can help questionable - since we're mucking around with it, now is a good time to make those changes.

[markspanbroek]

fwiw, a similar construct exists in the form of template valueOr, in Result itself

Yes, I've been able to make good use of valueOr in the |? operator.

as part of the general refresh happening there, there's a corresponding template errorOr as well now - also, lmk if there are any primitives needed in Result itself that can help questionable - since we're mucking around with it, now is a good time to make those changes.

Thanks, I appreciate that! There's one thing that I haven't been able to do, and that is provide a $ for Result[T, ref CatchableError]. If I remember correctly, it's related to a mixin $ that's somewhere in the Result library. I'll file an issue.

[markspanbroek]

If I remember correctly, it's related to a mixin $ that's somewhere in the Result library. I'll file an issue.

I investigated the issue a bit further; turns out it didn't have anything to do with the mixin, but was triggered in a roundabout way by other test code. I filed an issue: arnetheduck/nim-results#21

[dryajov]

Anything blocking this from being merged?

[markspanbroek]

Anything blocking this from being merged?

Not anymore, I found a nasty bug that needed fixing, but that's been dealt with in 15e514d. I'll make a new release shortly.