[phobos] Silent failure of std.container unittests

[Andrei Alexandrescu]

On 07/17/2010 12:07 AM, Walter Bright wrote:
>> But of course, that's probably more work to do in the compiler.
>>
>
> Yes, but it's not hard. For me, the issue is making things for the user
> more complicated. D is already a very large language, and adding more
> and more stuff like this makes it larger, buggier, and less
> approachable. For example, nothing stands out to say what "unittest
> assert" does vs "assert". I'd have to continuously recheck the manual.
> One thing I like about the current unittest setup is how utterly trivial
> it is to use effectively.

The way I understand what you say means that assert should have uniform 
semantics, whether used inside a unittest or not. Selective behavior 
would be one of those RTFM things.

> Is it really a good idea to have so much customizable behavior wedded
> into the grammar, whereas it can be achieved using existing D constructs
> (albeit with a little more typing)? I think the default behavior should
> be the simplest, "git 'er done" that covers 90% of the usage needs. The
> other 10% can be done with conventional D constructs.

I agree, and to me the natural consequent is - assert throws, period.


Andrei