is this a bug with writeln / std.algorithm.remove?

[Steven Schveighoffer]

On 2/16/18 8:51 AM, arturg wrote:
> On Friday, 16 February 2018 at 13:28:59 UTC, Steven Schveighoffer wrote:
>>
>> Strictly speaking, this is not necessarily proof that it's in phobos, 
>> there could have been changes elsewhere that cause one of the 
>> conditions to fail.
>>
>> However, testing this out, I found something very weird.
>>
>> If you pragma(msg, isInputRange!(typeof(dgs))); inside your file, then 
>> it compiles (and prints true for that pragma).
>>
>> Makes no sense at all. Definitely seems like a compiler bug. A 
>> pragma(msg) shouldn't affect the outcome.
>>
> 
> ok so what should the bug report look like?

You have a pretty good minimal test, put that in bugzilla along with the 
forum thread link and all the info we know. Mark it as a dmd bug, 
regression, along with the version where it regressed (2.076.1), and I 
would tag it as "rejects-valid"

-Steve