[Issue 16263] Error: template std.algorithm.iteration.permutations cannot deduce function from argument types !()(char[])

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16263

--- Comment #8 from Wyatt <wyatt.epp at gmail.com> ---
(In reply to Ketmar Dark from comment #6)
> not a bug within the current state of things

I'm afraid I disagree: things not behaving as documented is always a bug.

> not "WONTFIX", as it *may* be fixed if devs will decide to remove
> autodecoding. so there is little choice left -- as bugreport *is* invalid
> (not a bug at all). sadly, we don't have "NOTABUG" reason to close it.

Usually, we call that "a normal open bug"?  Maybe depending on whichever
tracking bug is "kill autodecoding forever like we should have years ago"? 
Just because a bug can't be resolved as-is (or even plausibly in the near term)
doesn't mean it's invalid or not a bug.

Or maybe the title is wrong and it should really about fixing the docs to give
some indication that this is a thing people are going to hit or fixing the
compiler to give a useful error message instead of what, to anyone sane, is an
utterly mystifying complaint that water isn't wet.  Or both.  Or hell, make it
work as expected without the ugly hack!  Is that really not on the table?  How
has the community become so desensitised to this problem that bugs like this
just get closed outright?

I already thought autodecoding was rubbish from other angles, but this is
somehow the first time I've hit it like this.  I can easily see how people
would get discouraged from D in general if std.algorithm, supposedly one of the
shining stars of Phobos and D idioms in general, manages to break horribly when
you use an ordinary array of one of the base types.  Jesus, what a dumpster
fire.

Oh yes, it's also worth noting that default BZ search, which probably 99% of
people will use before filing one (if they even bother), doesn't turn up closed
bugs, which makes this nonsense effectively invisible. I only found it because
I searched the NG first and found the automated message on D.issues because I
didn't limit it to D.learn. 

In other words, the only reason I didn't report this precise issue again only
five days later was because of an *accident*.

--