Hi Steve,
I've been playing with your idea and here are some results.
1. In edge.js, the JS function called from C# has a specific signature
and way of operating:
function callMeFromCSharp (object, callbackFunction) {
...
callbackFunction(error, result);
}
In your example, calling 'args.jqUnit.assert("hello");' doesn't work
because jqUnit.assert() doesn't have the right arguments.
(see: https://github.com/tjanczuk/edge#how-to-call-nodejs-from-c)
2. Calling the jqUnit.assertX() methods outside of a test context is
prohibited. Something like this is needed:
jqUnit.test(
"Blah",
function() {
jqUnit.assertTrue("Veritas AEterna", true);
}
);
3. Now, jqUnit.test() is close to the correct pattern for the
callMeFromCSharp() function. What's missing is the [error, result]
arguments in the callback second argument to jqUnit.test(). I'm
exploring that now.
On 2017-11-21 4:00 PM, Steve Grundell wrote:
> Oops, hit the "fire" button. It should have said this:
>
> jqUnit.asynTest("asd", function () {
>
> testThing({
>
> jqUnit: jqUnit,
>
> test: "Test1"
>
> })
>
> });
>
>
>
> On Tue, Nov 21, 2017 at 8:56 PM, Steve Grundell
> <[email protected] <mailto:[email protected]>>
> wrote:
>
> Let me explain my thinking, in code..
>
> c#:
>
> public something Invoke(dynamic args) {
>
> args.jqUnit.assert("hello");
>
> try {
>
> this.GetType().GetMEthod(args.test).Invoke(this, input);
>
> } catch (Exception e) {
>
> jqUnit.fail(e.toString());
>
> }
>
> }
>
> public void Test1() {
>
> // the test
>
> TheClass tc = new TheClass();
> int result = tc.ReturnOne();
>
> jqUnit.assertEquals("hi", 1, result);
>
> }
>
> node:
>
> var testThing = // the edge function
> jqUnit.asynTest("asd", function () {
>
> testThing({
>
> })
>
> })
>
>
> On Tue, Nov 21, 2017 at 2:59 PM, Joseph Scheuhammer
> <[email protected] <mailto:[email protected]>> wrote:
>
> On 2017-11-15 7:30 AM, Steve Grundell wrote:
> > ...
> >
> > But, why use a test framework for .NET at all - when you're
> already
> > inside jqUnit?
> >
> > Each jqUnit test can be a wrapper to call the C# test, and
> the jqUnit
> > object can be passed into C# at the start, so you can still do
> > jqUnit.assert.
>
> Thanks Steve.
>
> That's the technique used with our Linux code where I
> previously wrote
> that only the exported or interface functions from the C++
> add-on are
> tested. The way that is done is by calling the JavaScript
> version of
> the exported function code from within jqUnit, and then using
> assertX()
> on the return value. That matches exactly what you are proposing.
>
> A worry is when things go south inside the native code
> (C++/Linux, or
> C#/Windows). That happened with the Windows processes bridge
> where an
> IllegalArgumentException was thrown from within the C# code.
> When that
> happens, no result is coming back to jqUnit. How does one
> test for that
> using jqUnit?
>
> Or, those exceptions should be handled within the native code
> such that
> a result always comes back. I think. Hmmm... What about
> functions
> that don't return anything? Is that an issue?
>
> --
> ;;;;joseph.
>
> 'Call me hobophobic, but I don't think two vagrants should be
> allowed to marry.'
> - J. Tiedrich -
>
>
>
--
;;;;joseph.
'Call me hobophobic, but I don't think two vagrants should be allowed to marry.'
- J. Tiedrich -
_______________________________________________
Architecture mailing list
[email protected]
https://lists.gpii.net/mailman/listinfo/architecture
