Document that expected must be a RegExp

[domenic]

You get silent false positives if you use e.g. TypeError instead of /TypeError/.

[ghost]

Should it also accept a string?

[domenic]

I don't think so. Maybe it should accept constructors, but strings are not that great, especially given the message string adjacent to it in the parameter list.

[ljharb]

I'm currently using this by passing in TypeError, the constructor function. When I change it to use /^TypeError$/, I get a failure with expected: '/^TypeError$/', actual: { message: '…' }

Using /TypeError/ also passes, but that seems like it's abusing string coercion. Is this argument not expected to be the Error constructor function?

[domenic]

Look at the source code. /^TypeError$/ of course fails because the message is nonempty.

---

From: Jordan Harbandmailto:notifications@github.com
Sent: ý2014-ý09-ý16 23:48
To: substack/tapemailto:tape@noreply.github.com
Cc: Domenic Denicolamailto:domenic@domenicdenicola.com
Subject: Re: [tape] Document that expected must be a RegExp (#101)

I'm currently using this by passing in TypeError, the constructor function. When I change it to use /^TypeError$/, I get a failure with expected: '/^TypeError$/', actual: { message: '…' }

Using /TypeError/ also passes, but that seems like it's abusing string coercion. Is this argument not expected to be the Error constructor function?

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/101#issuecomment-55855323.

[ljharb]

So it seems like expected is totally unused unless it's a regex - that doesn't seem very useful to me :-/

@substack would you be willing to accept a PR that has an additional check for if expected is a function, and another if it's an Error object?

[domenic]

Seems fine to me. Overloading is an anti pattern.

On Sep 17, 2014, at 9:31, "Jordan Harband" <notifications@github.mirror.nvdadr.commailto:notifications@github.com> wrote:

So it seems like expected is totally unused unless it's a regex - that doesn't seem very useful to me :-/

@substackhttps://github.com/substack would you be willing to accept a PR that has an additional check for if expected is a function, and another if it's an Error object?

Reply to this email directly or view it on GitHubhttps://github.com//pull/101#issuecomment-55921246.

[Raynos]

@ljharb just open a PR.

Making test.throws() have the same semantics as require('assert').throws() is always a welcome change.