December 10, 2013
This morning I decided to use 2.0.64 (well actually, git HEAD) to recompile a project that's been on the backburner for a while, and at first it appeared as though a regression has occurred, as dmd spewed out a screenful of compile errors.

Upon closer inspection, though, the errors were caused by several if-conditions of the following sort:

	if (someCondition &&
	    cast(bool)(ptr = getPtr(...)) && // <--- compile error here
	    ...)
	{ ... }

These compile errors forced me to rewrite these lines as:

	if (someCondition &&
	    (ptr = getPtr(...)) !is null &&
	    ...)
	{ ... }

Which, I'm sure you'll agree, is far clearer in intent, and less prone to unexpected bugs.

So this isn't a regression; it's *pro*gression!

(And on that note, I'd like to propose that code breakage of this sort *should* be allowed in D. While we *should* be stabilizing the language, I don't think it's right to go to the opposite extreme of hindering language fixes just so badly-written code will continue to compile. Better to get user code cleaned up of similar unsafe practices than to allow backward compatibility to hold back D progress!)


T

-- 
What do you call optometrist jokes? Vitreous humor.
December 10, 2013
H. S. Teoh:

> So this isn't a regression; it's *pro*gression!
>
> (And on that note, I'd like to propose that code breakage of this sort
> *should* be allowed in D. While we *should* be stabilizing the language,
> I don't think it's right to go to the opposite extreme of hindering
> language fixes just so badly-written code will continue to compile.
> Better to get user code cleaned up of similar unsafe practices

I think this is another example of the same:
https://d.puremagic.com/issues/show_bug.cgi?id=4733

Bye,
bearophile