In https://dlang.org/phobos/std_exception.html#errnoEnforce this example is shown:

---
auto f = errnoEnforce(fopen("data.txt"));
auto line = readln(f);
enforce(line.length); // expect a non-empty line
---

I added

   import std.stdio;
   import std.exception;

and get an error message which reminds me of C++:

zz.d(8): Deprecation: std.stdio.fopen(R1, R2)(R1 name, R2 mode = "r") if ((isInputRange!R1 && isSomeChar!(ElementEncodingType!R1) || isSomeString!R1) && (isInputRange!R2 && isSomeChar!(ElementEncodingType!R2) || isSomeString!R2)) is not visible from module zz
zz.d(8): Error: function std.stdio.fopen!(string, string).fopen is not accessible from module zz
zz.d(9): Error: template std.stdio.readln cannot deduce function from argument types !()(shared(_IO_FILE)*), candidates are:
[...]/dmd2/linux/bin64/../../src/phobos/std/stdio.d(4068):        std.stdio.readln(S = string)(dchar terminator = '\x0a') if (isSomeString!S)
[...]/dmd2/linux/bin64/../../src/phobos/std/stdio.d(4102):        std.stdio.readln(C)(ref C[] buf, dchar terminator = '\x0a') if (isSomeChar!C && is(Unqual!C == C) && !is(C == enum))
[...]/dmd2/linux/bin64/../../src/phobos/std/stdio.d(4109):        std.stdio.readln(C, R)(ref C[] buf, R terminator) if (isSomeChar!C && is(Unqual!C == C) && !is(C == enum) && isBidirectionalRange!R && is(typeof(terminator.front == (dchar).init)))
make: *** [zz.o] Error 1

What's wrong here?

On Friday, June 29, 2018 01:25:39 kdevel via Digitalmars-d-learn wrote:
> In https://dlang.org/phobos/std_exception.html#errnoEnforce this example is shown:
>
> ---
> auto f = errnoEnforce(fopen("data.txt"));
> auto line = readln(f);
> enforce(line.length); // expect a non-empty line
> ---
>
> I added
>
>     import std.stdio;
>     import std.exception;
>
> and get an error message which reminds me of C++:
>
> zz.d(8): Deprecation: std.stdio.fopen(R1, R2)(R1 name, R2 mode =
> "r") if ((isInputRange!R1 && isSomeChar!(ElementEncodingType!R1)
>
> || isSomeString!R1) && (isInputRange!R2 &&
>
> isSomeChar!(ElementEncodingType!R2) || isSomeString!R2)) is not
> visible from module zz
> zz.d(8): Error: function std.stdio.fopen!(string, string).fopen
> is not accessible from module zz
> zz.d(9): Error: template std.stdio.readln cannot deduce function
> from argument types !()(shared(_IO_FILE)*), candidates are:
> [...]/dmd2/linux/bin64/../../src/phobos/std/stdio.d(4068):
> std.stdio.readln(S = string)(dchar terminator = '\x0a') if
> (isSomeString!S)
> [...]/dmd2/linux/bin64/../../src/phobos/std/stdio.d(4102):
> std.stdio.readln(C)(ref C[] buf, dchar terminator = '\x0a') if
> (isSomeChar!C && is(Unqual!C == C) && !is(C == enum))
> [...]/dmd2/linux/bin64/../../src/phobos/std/stdio.d(4109):
> std.stdio.readln(C, R)(ref C[] buf, R terminator) if
> (isSomeChar!C && is(Unqual!C == C) && !is(C == enum) &&
> isBidirectionalRange!R && is(typeof(terminator.front ==
> (dchar).init)))
> make: *** [zz.o] Error 1
>
> What's wrong here?

It looks like that example is using a private function from std.stdio which wraps core.stdc.stdio.fopen. It also passes a FILE* to readln, which is completely wrong. readln as a free function works with std.stdio.stdin and does not accept a FILE*, whereas readln on std.stdio.File obviously operates on File, since it's a member function on File. You could use core.stdc.stdio.fopen to get a FILE and pass it to File's constructor, and then call readln on the file, but really, that example needs to be completely redone.

The error message for fopen stems from the fact that it's private, and that message is longer than it would be otherwise because of an import bug that was previously fixed but generates a deprecation message for now so as not to immediately break code. The error messages after that have to do with the compiler showing you the various overloads of readln that don't match FILE*.

- Jonathan M Davis

On Friday, 29 June 2018 at 02:28:04 UTC, Jonathan M Davis wrote:
> [...] really, that example needs to be completely redone.

Shall I create a bug report?

On Friday, June 29, 2018 09:08:40 kdevel via Digitalmars-d-learn wrote:
> On Friday, 29 June 2018 at 02:28:04 UTC, Jonathan M Davis wrote:
> > [...] really, that example needs to be completely redone.
>
> Shall I create a bug report?

Yes. Aside from someone trying it out and complaining about it, it probably wouldn't be noticed or fixed, since it's one of the few tests that doesn't work as a ddoc-ed unit test.

- Jonathan M Davis

On Friday, 29 June 2018 at 09:22:03 UTC, Jonathan M Davis wrote:
>> Shall I create a bug report?
>
> Yes. Aside from someone trying it out and complaining about it, it probably wouldn't be noticed or fixed, since it's one of the few tests that doesn't work as a ddoc-ed unit test.

Issue 19041 - errnoEnforce: example does not compile