digitalmars.D.learn - is it bug?
- drug (23/23) Jan 30 2015 ```
- drug (2/2) Jan 30 2015 The real problem is if I comment my unittest out then it compiles. Why
- anonymous (11/35) Jan 30 2015 Reduced:
- anonymous (3/9) Jan 30 2015 (Tab + Enter strikes again.)
- BBaz (12/18) Jan 30 2015 have you tried
```
import std.container: RedBlackTree;
class Manager(TT, alias Cmp = "a<b")
{
alias Container = RedBlackTree!(TT, Cmp);
Container _cont;
static init()
{
auto instance = new typeof(this)();
instance._cont = new Container();
return instance;
}
}
unittest
{
alias IntManager = Manager!int;
auto im = IntManager.init();
}
void main() {}
```
run it using
rdmd -unittest filename.d
it fails in a RedBlackTree unittest with 2.065 and 2.066, is it bug?
Jan 30 2015
drug <drug2004 bk.ru> The real problem is if I comment my unittest out then it compiles. Why my unittest causes this behaviour?
Jan 30 2015
On Friday, 30 January 2015 at 12:32:05 UTC, drug wrote:
```
import std.container: RedBlackTree;
class Manager(TT, alias Cmp = "a<b")
{
alias Container = RedBlackTree!(TT, Cmp);
Container _cont;
static init()
{
auto instance = new typeof(this)();
instance._cont = new Container();
return instance;
}
}
unittest
{
alias IntManager = Manager!int;
auto im = IntManager.init();
}
void main() {}
```
run it using
rdmd -unittest filename.d
it fails in a RedBlackTree unittest with 2.065 and 2.066, is it
bug?
Reduced:
import std.container: RedBlackTree;
alias Container = RedBlackTree!(int, "a<b");
Error: [...]/std/container/rbtree.d(850): unittest failure
Lines 846-850:
static if(less == "a < b")
auto vals = [1, 2, 3, 4, 5];
else
auto vals = [5, 4, 3, 2, 1];
assert(equal(r, vals));
Jan 30 2015
On Friday, 30 January 2015 at 13:11:35 UTC, anonymous wrote:
Lines 846-850:
static if(less == "a < b")
auto vals = [1, 2, 3, 4, 5];
else
auto vals = [5, 4, 3, 2, 1];
assert(equal(r, vals));
(Tab + Enter strikes again.)
That test looks wrong. So, Phobos bug I think.
Jan 30 2015
On 30.01.2015 16:14, anonymous wrote:On Friday, 30 January 2015 at 13:11:35 UTC, anonymous wrote:I guess it is bug in this unittest, because it fails if I set predicate less to "a<b" instead of "a < b". Spaces rule in this case.Lines 846-850: static if(less == "a < b") auto vals = [1, 2, 3, 4, 5]; else auto vals = [5, 4, 3, 2, 1]; assert(equal(r, vals));(Tab + Enter strikes again.) That test looks wrong. So, Phobos bug I think.
Jan 30 2015
On 30.01.2015 16:35, drug wrote:On 30.01.2015 16:14, anonymous wrote:Filed a bug https://issues.dlang.org/show_bug.cgi?id=14082On Friday, 30 January 2015 at 13:11:35 UTC, anonymous wrote:I guess it is bug in this unittest, because it fails if I set predicate less to "a<b" instead of "a < b". Spaces rule in this case.Lines 846-850: static if(less == "a < b") auto vals = [1, 2, 3, 4, 5]; else auto vals = [5, 4, 3, 2, 1]; assert(equal(r, vals));(Tab + Enter strikes again.) That test looks wrong. So, Phobos bug I think.
Jan 30 2015
On Friday, 30 January 2015 at 12:32:05 UTC, drug wrote:
static init()
{
auto instance = new typeof(this)();
instance._cont = new Container();
return instance;
}
have you tried
---
static typeof(this) init()
{
auto instance = new typeof(this)();
instance._cont = new Container();
return instance;
}
---
?
because currently either "void" or a type is missing.
Jan 30 2015
On 30.01.2015 16:31, BBaz wrote:On Friday, 30 January 2015 at 12:32:05 UTC, drug wrote:Type is inferred automatically.static init() { auto instance = new typeof(this)(); instance._cont = new Container(); return instance; }have you tried --- static typeof(this) init() { auto instance = new typeof(this)(); instance._cont = new Container(); return instance; } --- ? because currently either "void" or a type is missing.
Jan 30 2015
On Friday, 30 January 2015 at 13:34:57 UTC, drug wrote:On 30.01.2015 16:31, BBaz wrote:right, i've forgot that init() is a >>built-in<< property. But it seemsthat i'm not the only one...On Friday, 30 January 2015 at 12:32:05 UTC, drug wrote:Type is inferred automatically.static init() { auto instance = new typeof(this)(); instance._cont = new Container(); return instance; }have you tried --- static typeof(this) init() { auto instance = new typeof(this)(); instance._cont = new Container(); return instance; } --- ? because currently either "void" or a type is missing.
Jan 30 2015
On Friday, 30 January 2015 at 13:39:05 UTC, BBaz wrote:On Friday, 30 January 2015 at 13:34:57 UTC, drug wrote:I think that doesn't matter here. Still works fine if you name it something other than "init". I don't know if this works as designed, or if an "auto" return type should be required. Besides, it's a bad idea to call a member "init", because it steals the name of the default initializer. It doesn't override the default initializer, it just takes its name. The compiler should not accept it, in my opinion.On 30.01.2015 16:31, BBaz wrote:right, i've forgot that init() is a >>built-in<< property. But it seemsthat i'm not the only one...On Friday, 30 January 2015 at 12:32:05 UTC, drug wrote:Type is inferred automatically.static init() { auto instance = new typeof(this)(); instance._cont = new Container(); return instance; }have you tried --- static typeof(this) init() { auto instance = new typeof(this)(); instance._cont = new Container(); return instance; } --- ? because currently either "void" or a type is missing.
Jan 30 2015
On Friday, 30 January 2015 at 13:56:17 UTC, anonymous wrote:On Friday, 30 January 2015 at 13:39:05 UTC, BBaz wrote:Yes, that was the point: "bad idea to call a member "init". But I've missed something with inference of return type... let's call this the "BMS" : big-mouth-syndrom... I didn't know it could be ommitted...I thought it could be if the function has the' safe'attribute... Sorry drug, BMS...On Friday, 30 January 2015 at 13:34:57 UTC, drug wrote:I think that doesn't matter here. Still works fine if you name it something other than "init". I don't know if this works as designed, or if an "auto" return type should be required. Besides, it's a bad idea to call a member "init", because it steals the name of the default initializer. It doesn't override the default initializer, it just takes its name. The compiler should not accept it, in my opinion.On 30.01.2015 16:31, BBaz wrote:right, i've forgot that init() is a >>built-in<< property. But it seemsthat i'm not the only one...On Friday, 30 January 2015 at 12:32:05 UTC, drug wrote:Type is inferred automatically.static init() { auto instance = new typeof(this)(); instance._cont = new Container(); return instance; }have you tried --- static typeof(this) init() { auto instance = new typeof(this)(); instance._cont = new Container(); return instance; } --- ? because currently either "void" or a type is missing.
Jan 30 2015
On 30.01.2015 17:04, BBaz wrote:Yes, that was the point: "bad idea to call a member "init". But I've missed something with inference of return type... let's call this the "BMS" : big-mouth-syndrom... I didn't know it could be ommitted...I thought it could be if the function has the' safe'attribute...static helps to infer typeSorry drug, BMS...there is nothing to sorry
Jan 30 2015
On 30.01.2015 16:56, anonymous wrote:Besides, it's a bad idea to call a member "init", because it steals the name of the default initializer. It doesn't override the default initializer, it just takes its name. The compiler should not accept it, in my opinion.Good remark! I'll rename it.
Jan 30 2015