From: knu@...
Date: 2016-11-05T13:54:04+00:00
Subject: [ruby-core:77972] [Ruby trunk Bug#12799] IPAddr#== throws an exception with objects that can't be converted to IPAddr

Issue #12799 has been updated by Akinori MUSHA.


Committed, thanks!

----------------------------------------
Bug #12799: IPAddr#== throws an exception with objects that can't be converted to IPAddr
https://bugs.ruby-lang.org/issues/12799#change-61306

* Author: Hiroshi SHIBATA
* Status: Closed
* Priority: Normal
* Assignee: Akinori MUSHA
* ruby -v: 
* Backport: 2.1: UNKNOWN, 2.2: UNKNOWN, 2.3: UNKNOWN
----------------------------------------
I accidentally deleted issue #12779. It was created by https://bugs.ruby-lang.org/users/11301

----

Description

IPAddr#== uses IPAddr#coerce_other to attempt to convert the 'other' object into an IPAddr for comparison.
If this conversion fails, an exception is thrown, either 'IPAddr::InvalidAddressError' or a 'NoMethodError' referring to the lack of a 'to_i' function.

I would expect a call to == should only return a true or a false in any circumstances. If the tested object is not an IPAddr then clearly it should be false.

```
IPAddr.new("1.1.1.1")=="sometext"
raises IPAddr::InvalidAddressError: invalid address
"sometext"==IPAddr.new("1.1.1.1")
```

returns false.

This also causes other core functions to fail..

```
ip = IPAddr.new("1.1.1.1")
iprange = IPAddr.new("2.2.2.2") .. IPAddr.new("2.2.2.20")
arr = [ip, iprange]

arr.delete(iprange)
```

Results in NoMethodError: undefined method `to_i' for #Range:0x00000001901748

I have patched this at github https://github.com/ruby/ruby/pull/1435

and produced a diff file from the above pull request, apologies if I have done so incorrectly

----

#1[ruby-core:77365] Updated by Akinori MUSHA 5 days ago

Good point. Would you mind adding a regression test to test/test_ipaddr.rb?

----

#2[ruby-core:77387] Updated by David Sexton 4 days ago

Akinori MUSHA wrote:

>Good point. Would you mind adding a regression test to test/test_ipaddr.rb?

No problem.

I have also noticed the same issue with <=> so have added a patch (and test) for that to return nil

Updates diff is at https://github.com/ruby/ruby/pull/1435.diff





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

Unsubscribe: <mailto:ruby-core-request@ruby-lang.org?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>