Bug #2945: Regexp#=== is failed by an exception when the exception is occurred in method_missing

• Status: Rejected
• Assignee: -
• Target version: 1.9.2
• ruby -v: ruby 1.9.2dev (2010-03-08 trunk 26849) [x86_64-darwin10.2.0]

[mrkn (Kenta Murata)]

=begin
When an exception is occurred in method_missing,
Regexp#=== is failed by the exception.
Is this valid behavior?

o = Object.new
class <<
def method_missing(m, *args)
raise "XXX" if m == :to_str
end
end
/aaa/ === o # =:> `method_missing': XXX (RuntimeError)

If correct this, apply the following patch.

diff --git a/re.c b/re.c
index 6ec3646..8a38155 100644
--- a/re.c
+++ b/re.c
@@ -2685,6 +2685,12 @@ rb_reg_match(VALUE re, VALUE str)
return LONG2FIX(pos);
}

+static VALUE
+reg_check_operand(VALUE str)
+{

• return reg_operand(str, FALSE);
  +}

/*

• call-seq:

• rxp === str   => true or false
  

@@ -2708,7 +2714,7 @@ rb_reg_eqq(VALUE re, VALUE str)
{
long start;

• str = reg_operand(str, FALSE);

• str = rb_protect(reg_check_operand, str, 0);
  if (NIL_P(str)) {
  rb_backref_set(Qnil);
  return Qfalse;
  =end

[matz (Yukihiro Matsumoto)]

=begin
Hi,

In message "Re: [ruby-core:28552] [Bug #2945] Regexp#=== is failed by an exception when the exception is occurred in method_missing"
on Mon, 8 Mar 2010 15:32:24 +0900, Kenta Murata [e-mail] writes:

|When an exception is occurred in method_missing,
|Regexp#=== is failed by the exception.
|Is this valid behavior?

I think so. What makes you think it's invalid?

						matz.

=end

[mrkn (Kenta Murata)]

=begin
Hi,

On 2010/03/08, at 15:46, Yukihiro Matsumoto wrote:

Hi,

In message "Re: [ruby-core:28552] [Bug #2945] Regexp#=== is failed by an exception when the exception is occurred in method_missing"
on Mon, 8 Mar 2010 15:32:24 +0900, Kenta Murata [e-mail] writes:

|When an exception is occurred in method_missing,
|Regexp#=== is failed by the exception.
|Is this valid behavior?

I think so. What makes you think it's invalid?

I think the present behavior is unfortunate the following case.
I don't expect raising exception, but printing 2.

o1 = Object.new
class << o1
def method_missing(*args)
raise "XXX"
end
end

case o1
when /aaa/
puts 1
when o
puts 2
end
#=> `method_missing': XXX (RuntimeError)

--
Kenta Murata
OpenPGP FP = FA26 35D7 4F98 3498 0810 E0D5 F213 966F E9EB 0BCC

E-mail: [e-mail]
twitter: http://twitter.com/mrkn/
blog: http://d.hatena.ne.jp/mrkn/

=end

[matz (Yukihiro Matsumoto)]

=begin
Hi,

In message "Re: [ruby-core:28554] Re: [Bug #2945] Regexp#=== is failed by an exception when the exception is occurred in method_missing"
on Mon, 8 Mar 2010 16:56:14 +0900, Kenta Murata [e-mail] writes:

|> |Is this valid behavior?
|>
|> I think so. What makes you think it's invalid?
|
|I think the present behavior is unfortunate the following case.
|I don't expect raising exception, but printing 2.

I have to repeat myself. Why do you expect so?
#method_missing raised an exception, no one captured. Why do we
expect exceptions to be captured implicitly?

| o1 = Object.new
| class << o1
| def method_missing(*args)
| raise "XXX"
| end
| end
|
| case o1
| when /aaa/
| puts 1
| when o
| puts 2
| end
| #=> `method_missing': XXX (RuntimeError)
|
|--
|Kenta Murata
|OpenPGP FP = FA26 35D7 4F98 3498 0810 E0D5 F213 966F E9EB 0BCC
|
|E-mail: [e-mail]
|twitter: http://twitter.com/mrkn/
|blog: http://d.hatena.ne.jp/mrkn/
|
|

=end

[mrkn (Kenta Murata)]

=begin
Hi,

On 2010/03/08, at 17:12, Yukihiro Matsumoto wrote:

|> |Is this valid behavior?
|>
|> I think so. What makes you think it's invalid?
|
|I think the present behavior is unfortunate the following case.
|I don't expect raising exception, but printing 2.

I have to repeat myself. Why do you expect so?
#method_missing raised an exception, no one captured. Why do we
expect exceptions to be captured implicitly?

Ok, I agree with you about that exceptions, no one captured,
shouldn't be captured implicitly.

But, I do still not understand Regexp#=== depends on method_missing.
I expect Regexp#=== with a non-string argument returns false.
In Regexp#===, if an argument doesn't have to_str method,
the argument is interpreted as a non-string, isn't it?
If so, Regexp#=== shouldn't raise an exception, but return false,
independently of the definition of method_missing.

| o1 = Object.new
| class << o1
| def method_missing(*args)
| raise "XXX"
| end
| end
|
| case o1
| when /aaa/
| puts 1
| when o
| puts 2
| end
| #=> `method_missing': XXX (RuntimeError)

--
Kenta Murata
OpenPGP FP = FA26 35D7 4F98 3498 0810 E0D5 F213 966F E9EB 0BCC

本を書きました!!
『Ruby 逆引きレシピ』 http://www.amazon.co.jp/dp/4798119881/mrkn-22

E-mail: [e-mail]
twitter: http://twitter.com/mrkn/
blog: http://d.hatena.ne.jp/mrkn/

=end

[matz (Yukihiro Matsumoto)]

=begin
Hi,

In message "Re: [ruby-core:28556] Re: [Bug #2945] Regexp#=== is failed by an exception when the exception is occurred in method_missing"
on Mon, 8 Mar 2010 17:47:26 +0900, Kenta Murata [e-mail] writes:

|But, I do still not understand Regexp#=== depends on method_missing.
|I expect Regexp#=== with a non-string argument returns false.
|In Regexp#===, if an argument doesn't have to_str method,
|the argument is interpreted as a non-string, isn't it?

It's a wrong design of method_missing. If one want exceptions from
method_missing ignored for implicit type conversion (to_str, etc.),
it should raise NoMethodError.

						matz.

=end

[mrkn (Kenta Murata)]

=begin
Hi,

On 2010/03/08, at 17:56, Yukihiro Matsumoto wrote:

|But, I do still not understand Regexp#=== depends on method_missing.
|I expect Regexp#=== with a non-string argument returns false.
|In Regexp#===, if an argument doesn't have to_str method,
|the argument is interpreted as a non-string, isn't it?

It's a wrong design of method_missing. If one want exceptions from
method_missing ignored for implicit type conversion (to_str, etc.),
it should raise NoMethodError.

I understand, I'll use NoMethodError for implicit type conversion methods
in method_missing.

Thanks.

--
Kenta Murata
OpenPGP FP = FA26 35D7 4F98 3498 0810 E0D5 F213 966F E9EB 0BCC

本を書きました!!
『Ruby 逆引きレシピ』 http://www.amazon.co.jp/dp/4798119881/mrkn-22

E-mail: [e-mail]
twitter: http://twitter.com/mrkn/
blog: http://d.hatena.ne.jp/mrkn/

=end

[mrkn (Kenta Murata)]

• Status changed from Open to Rejected

=begin

=end