Make assert work in oldIE

[domenic]

The assert interface depends on the expect and should interfaces. This is a real shame because it stops us from working in IE<=8. We should make it work!

[logicalparadox]

I agree, though admittedly, IE8 wasn't what put this topic on my radar initially. Unfortunately, the assert interface has always gotten less attention than expect and should interfaces because it was an afterthought. As such, all the plugins that have come out are "chainables". If we can work to fully separate the "core assertions" from the interfaces, and do so in such a way that they are "way-back" browser compatible, we can accomplish this IE requirement and make it easier for plugins to expose their features to multiple interfaces. Win-win.

I want to do this right, and a lot of internals will need to be shifted to accomplish this.

[rowanmanning]

I got burned by this the other day actually, glad to see it's been raised. I'm not familiar with the code-base (so I may be more of a hindrance than anything) but I'd be willing to help out with this if you need it.

[tmcw]

Current issues here are the use of Object.create (for which there's an easy polyfill) and Object.defineProperty (for which a polyfill is trickier to pull off)

[logicalparadox]

Discussion of how to refactor the Plugin API for more flexibility is taking place on this google group thread. If implemented correctly, a glorious side-effect of this refactor would be an assert interface that is free from needing defineProperty support and could work in legacy IE.

Please head over to the Google Group and read up/discuss!

[nzakas]

Just curious if this is still forthcoming or not? Lack of oldIE support is the big things preventing us from adopting Chai.

[jokeyrhyme]

I have a solution that is working for me in IE8: https://github.com/jokeyrhyme/assertive-chai.js
I've just ripped off the Assert API, as it's the fancy Expect / Should hooks that seem to be where much of the ES5 is mandated.

[mightyiam]

+1