std.unittests for (final?) review

[Jonathan M Davis]

On Saturday 08 January 2011 15:21:44 Jim wrote:
> I have to agree.
> 
> assert(1 + 1 < 3);
> 
> will always be easier to read than any:
> 
> assertPred!"a < b"(1 + 1, 3);
> 
> Why not just keep it simple and straight forward? Anything expressed in the
> latter form can be expressed in the former.

Well, honestly, after seeing enough stuff like the second example, I have no 
trouble reading it pretty much as quickly as the first one - especially with more 
complex expressions -  but I can see why some would consider that an issue.

However, my take on this is that the built-in assert cannot currently give us 
the kind of error messages that assertPred!() will. So, regardless of whether 
the built-in assert is improved later, if we want that kind of functionality 
now, then we need to have something like assertPred!().

Now, it would be extremely cool if the built in assert were able to give us 
error messages similar to what assertPred!() will do, so it could very well be 
valuable to add that later, but it's an implementation detail (similar to 
improving error messages), it's not necessarily going to be very easy to do, and 
assertPred!() will have already given us the necessary functionality, so I 
wouldn't expect such a change to be particularly high priority. It would 
probably have to be done by someone other than Walter who had the time and 
interest in doing it, since Walter has plenty of more important things an his 
plate.

So, assert my eventually be that cool, but it isn't now, and it may never be, so 
we create assertPred!() now, and then we have a solution. If assert eventually 
_is_ that cool, then it'll be an option to use that instead of assertPred!() for 
getting better error messages. But for now, it isn't an option.

- Jonathan M Davis