[ruby-core:107521] [Ruby master Bug#18577] Range#include? returns wrong result for beginless range with exclusive string end

[ruby-core:107521] [Ruby master Bug#18577] Range#include? returns wrong result for beginless range with exclusive string end

[takaram (Takuya Aramaki)]

Issue #18577 has been reported by takaram (Takuya Aramaki).

----------------------------------------
Bug #18577: Range#include? returns wrong result for beginless range with exclusive string end
https://bugs.ruby-lang.org/issues/18577

* Author: takaram (Takuya Aramaki)
* Status: Open
* Priority: Normal
* ruby -v: ruby 3.1.0p0 (2021-12-25 revision fb4df44d16) [x86_64-linux]
* Backport: 2.6: UNKNOWN, 2.7: UNKNOWN, 3.0: UNKNOWN, 3.1: UNKNOWN
----------------------------------------
Exclusive range should not include the end value, but `(...'z').include? 'z'` returns true. `member?` and `===` behave in the same way, while `cover?` does not.
This happens when the range is beginless and its end is string.

```
irb(main):001:0> (...'z').include? 'z'
=> true
irb(main):002:0> (...'z').member? 'z'
=> true
irb(main):003:0> (...'z') === 'z'
=> true
irb(main):004:0> (...'z').cover? 'z'
=> false
```



-- 
https://bugs.ruby-lang.org/

[ruby-core:107529] [Ruby master Bug#18577] Range#include? returns wrong result for beginless range with exclusive string end

[zverok (Victor Shepelev)]

Issue #18577 has been updated by zverok (Victor Shepelev).


It seems that the cause is unnecessary (?) [specialization](https://github.com/ruby/ruby/blob/master/range.c#L1782) of `Range#include?` for String arguments.

I believe string ranges should just fallback to the default implementation:
```ruby
(...Date.today) === Date.today
# => false 
(...Date.today).cover? Date.today
# => false 
(...Date.today).include? Date.today
# in `each': can't iterate from NilClass (TypeError)
```
The removal of specialization will also fix current (and possible future) discrepancies of `Range#cover?` and `Range#===` for String ranges.

If it is not acceptable (some cases of backward compatibility?), [this statement](https://github.com/ruby/ruby/blob/master/range.c#L1795) should be fixed.

cc @jeremyevans0

----------------------------------------
Bug #18577: Range#include? returns wrong result for beginless range with exclusive string end
https://bugs.ruby-lang.org/issues/18577#change-96440

* Author: takaram (Takuya Aramaki)
* Status: Open
* Priority: Normal
* ruby -v: ruby 3.1.0p0 (2021-12-25 revision fb4df44d16) [x86_64-linux]
* Backport: 2.6: UNKNOWN, 2.7: UNKNOWN, 3.0: UNKNOWN, 3.1: UNKNOWN
----------------------------------------
Exclusive range should not include the end value, but `(...'z').include? 'z'` returns true. `member?` and `===` behave in the same way, while `cover?` does not.
This happens when the range is beginless and its end is string.

```
irb(main):001:0> (...'z').include? 'z'
=> true
irb(main):002:0> (...'z').member? 'z'
=> true
irb(main):003:0> (...'z') === 'z'
=> true
irb(main):004:0> (...'z').cover? 'z'
=> false
```



-- 
https://bugs.ruby-lang.org/

[ruby-core:107534] [Ruby master Bug#18577] Range#include? returns wrong result for beginless range with exclusive string end

[jeremyevans0 (Jeremy Evans)]

Issue #18577 has been updated by jeremyevans0 (Jeremy Evans).


zverok (Victor Shepelev) wrote in #note-1:
> It seems that the cause is unnecessary (?) [specialization](https://github.com/ruby/ruby/blob/master/range.c#L1782) of `Range#include?` for String arguments.
> 
> I believe string ranges should just fallback to the default implementation:
> ```ruby
> (...Date.today) === Date.today
> # => false 
> (...Date.today).cover? Date.today
> # => false 
> (...Date.today).include? Date.today
> # in `each': can't iterate from NilClass (TypeError)
> ```
> The removal of specialization will also fix current (and possible future) discrepancies of `Range#cover?` and `Range#===` for String ranges.

If you want to remove the specialization, that's something that probably should be discussed at a developer meeting.  I am against removing it.  Not because I think the specialization is a good idea, but because I don't think it's worth breaking backwards compatibility (raising TypeError for previously working code).

> If it is not acceptable (some cases of backward compatibility?), [this statement](https://github.com/ruby/ruby/blob/master/range.c#L1795) should be fixed.

I submitted a pull request to fix this: https://github.com/ruby/ruby/pull/5541

----------------------------------------
Bug #18577: Range#include? returns wrong result for beginless range with exclusive string end
https://bugs.ruby-lang.org/issues/18577#change-96446

* Author: takaram (Takuya Aramaki)
* Status: Open
* Priority: Normal
* ruby -v: ruby 3.1.0p0 (2021-12-25 revision fb4df44d16) [x86_64-linux]
* Backport: 2.6: UNKNOWN, 2.7: UNKNOWN, 3.0: UNKNOWN, 3.1: UNKNOWN
----------------------------------------
Exclusive range should not include the end value, but `(...'z').include? 'z'` returns true. `member?` and `===` behave in the same way, while `cover?` does not.
This happens when the range is beginless and its end is string.

```
irb(main):001:0> (...'z').include? 'z'
=> true
irb(main):002:0> (...'z').member? 'z'
=> true
irb(main):003:0> (...'z') === 'z'
=> true
irb(main):004:0> (...'z').cover? 'z'
=> false
```



-- 
https://bugs.ruby-lang.org/