# Coerce(), what protocol to implement it

**URL:** <https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291>\
**Category:** ruby-talk\
**Created:** [29 April 2004 18:45 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291 "2004-04-29T18:45:40Z")\
**Posts on this page:** 14\
**Page:** 1

<div class="post-metadata">

**Author:** ![Jean-Hugues\_ROBERT](https://avatars.discourse-cdn.com/v4/letter/j/e47c2d/32.png) [@Jean-Hugues\_ROBERT](https://rubytalk.org/u/Jean-Hugues_ROBERT)\
**Post date:** [29 April 2004 18:45 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/1 "2004-04-29T18:45:40Z")

</div>

Hi,

I need to write a class whose instances can be on the left side of  
a + operator. I am ready to implement whatever make the other side  
happy. But what is that I need to implement ?

I am trying to understand coerce()

Apparently when a receiver is not happy with the type of a parameter,  
it invokes coerce() on the parameter.  
if not\_happy\_with? param then x = param.coerce( self)

coerce() then returns an array of two items. Both items can be new  
objects. However coerce() says that they are “compatible”.

Question:

1. What happens if coerce() cannot produce such a result when there  
are no such “compatible” objects ? Apparently coerce() raises an  
ArgumentError exception. Right ?

2. How can I make MyClass instances coercible ? That must be possible,  
because 1.coerce( “2”) works, yet “2”.coerce( 1) does not…

3. Why does 1.coerce( “3”) returns [3.0,1.0] instead of [1,3] while  
1.coerce( 3) returns [1,3] ?

4. Can I redefine coerce() ? If so, when is it called ? (I tried,  
it is not called…)

5. By defining to\_str() I can have “xx” + MyClass.new(xx) work. What  
do I have to redefine to have 1 + MyClass.new(xx) work ?

Google “ruby coerce” provided little help.

Thanks in advance,

Yours,

Jean-Hugues

> **···**
>
> * * *
> 
> Web: [http://hdl.handle.net/1030.37/1.1](http://hdl.handle.net/1030.37/1.1)  
> Phone: +33 (0) 4 92 27 74 17

---

<div class="post-metadata">

**Author:** ![Jean-Hugues\_ROBERT](https://avatars.discourse-cdn.com/v4/letter/j/e47c2d/32.png) [@Jean-Hugues\_ROBERT](https://rubytalk.org/u/Jean-Hugues_ROBERT)\
**Post date:** [29 April 2004 19:30 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/2 "2004-04-29T19:30:09Z")

</div>

> Hi,
> 
> I need to write a class whose instances can be on the left side of

Sorry, I meant on the right side. 38 and still I can figure it out 😉

> **···**
>
> At 03:45 30/04/2004 +0900, you wrote:
> 
> > a + operator. I am ready to implement whatever make the other side  
> > happy. But what is that I need to implement ?
> > 
> > I am trying to understand coerce()
> > 
> > Apparently when a receiver is not happy with the type of a parameter,  
> > it invokes coerce() on the parameter.  
> > if not\_happy\_with? param then x = param.coerce( self)
> > 
> > coerce() then returns an array of two items. Both items can be new  
> > objects. However coerce() says that they are “compatible”.
> > 
> > Question:
> > 
> > 1. What happens if coerce() cannot produce such a result when there  
> > are no such “compatible” objects ? Apparently coerce() raises an  
> > ArgumentError exception. Right ?
> > 
> > 2. How can I make MyClass instances coercible ? That must be possible,  
> > because 1.coerce( “2”) works, yet “2”.coerce( 1) does not…
> > 
> > 3. Why does 1.coerce( “3”) returns [3.0,1.0] instead of [1,3] while  
> > 1.coerce( 3) returns [1,3] ?
> > 
> > 4. Can I redefine coerce() ? If so, when is it called ? (I tried,  
> > it is not called…)
> > 
> > 5. By defining to\_str() I can have “xx” + MyClass.new(xx) work. What  
> > do I have to redefine to have 1 + MyClass.new(xx) work ?
> > 
> > Google “ruby coerce” provided little help.
> > 
> > Thanks in advance,
> > 
> > Yours,
> > 
> > Jean-Hugues
> > 
> > * * *
> > 
> > Web: [@jhr is virteal, virtually real](http://hdl.handle.net/1030.37/1.1)  
> > Phone: +33 (0) 4 92 27 74 17
> 
> * * *
> 
> Web: [@jhr is virteal, virtually real](http://hdl.handle.net/1030.37/1.1)  
> Phone: +33 (0) 4 92 27 74 17

---

<div class="post-metadata">

**Author:** ![Christoph](https://avatars.discourse-cdn.com/v4/letter/c/eb8c5e/32.png) [@Christoph](https://rubytalk.org/u/Christoph)\
**Post date:** [29 April 2004 23:39 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/3 "2004-04-29T23:39:02Z")

</div>

Jean-Hugues ROBERT wrote:  
…

> Google “ruby coerce” provided little help.
> 
> Thanks in advance,

The coerce framework makes up for the lack of  
multiple dispatch aka “method overloading”,  
whose lack can be a major nuisance when doing  
numerical work. You probably should study the  
source code of the standard libraries “complex”  
or "rational, if you want to get a feel for  
using this framework. Here is a scaled down  
example

> **···**
>
> * * *
> 
> class MyClass
> 
> def initialize(val)  
> @value = val  
> end
> 
> def coerce(other)  
> [MyClass.new(other),self]  
> end
> 
> def -(lhs)  
> if MyClass === lhs  
> MyClass::new(@value - lhs.value)  
> else  
> x,y = lhs.coerce(@value)  
> MyClass::new(x - y)  
> end  
> end
> 
> protected  
> attr\_reader :value  
> end
> 
> p 1 - MyClass.new(5)  
> p MyClass.new(5) - 1
> 
> require ‘complex’
> 
> z = Complex(1.0,1)
> 
> ## p z - MyClass.new(5) p MyClass.new(5) - z
> 
> ## #\<MyClass:0x28a63e8 @value=-4\> #\<MyClass:0x28a6358 @value=4\> #\<MyClass:0x28955b8 @value=Complex(-4.0, 1)\> #\<MyClass:0x28954e0 @value=Complex(4.0, -1)\>
> 
> /Christoph

---

<div class="post-metadata">

**Author:** ![Robert](https://avatars.discourse-cdn.com/v4/letter/r/e95f7d/32.png) [@Robert](https://rubytalk.org/u/Robert)\
**Post date:** [30 April 2004 07:54 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/4 "2004-04-30T07:54:02Z")

</div>

“Jean-Hugues ROBERT” [jean\_hugues\_robert@yahoo.com](mailto:jean_hugues_robert@yahoo.com) schrieb im Newsbeitrag  
[news:6.0.1.1.0.20040429190535.01cd1ef0@pop.mail.yahoo.com](mailto:news:6.0.1.1.0.20040429190535.01cd1ef0@pop.mail.yahoo.com)…

> Hi,
> 
> I need to write a class whose instances can be on the left side of  
> a + operator. I am ready to implement whatever make the other side  
> happy. But what is that I need to implement ?
> 
> I am trying to understand coerce()
> 
> Apparently when a receiver is not happy with the type of a parameter,  
> it invokes coerce() on the parameter.  
> if not\_happy\_with? param then x = param.coerce( self)
> 
> coerce() then returns an array of two items. Both items can be new  
> objects. However coerce() says that they are “compatible”.
> 
> Question:
> 
> 1. What happens if coerce() cannot produce such a result when there  
> are no such “compatible” objects ? Apparently coerce() raises an  
> ArgumentError exception. Right ?
> 
> 2. How can I make MyClass instances coercible ? That must be possible,  
> because 1.coerce( “2”) works, yet “2”.coerce( 1) does not…
> 
> 3. Why does 1.coerce( “3”) returns [3.0,1.0] instead of [1,3] while  
> 1.coerce( 3) returns [1,3] ?

Just a minor note: this is not true:

irb(main):017:0\> 1.coerce 3  
=\> [3, 1]

Order matters! coerce always returns values in reverse order of the  
source.

> 1. Can I redefine coerce() ? If so, when is it called ? (I tried,  
> it is not called…)
> 
> 2. By defining to\_str() I can have “xx” + MyClass.new(xx) work. What  
> do I have to redefine to have 1 + MyClass.new(xx) work ?

It depends: there are two basic options, which are important to  
understand:

i) You class can easily be converted to a standard numerical type. In  
that case you can use the standard implementation of operators:

class YourClass  
def to\_i; 1; end  
def coerce(x) [x, to\_i]; end  
def +(o); o + self; end  
end

irb(main):034:0\> 1 + YourClass.new  
=\> 2  
irb(main):035:0\> YourClass.new + 1  
=\> 2

ii) There is no such easy conversion then you probably need to convert  
other types to your class:

class YourClass  
def initialize(x=“foo”); @x = x; end  
def coerce(x) [self.class.new( x ), self]; end  
def +(o); self.class.new(“addition(#{self.inspect}, #{o.inspect})”); end  
def -(o); self.class.new(“substraction(#{self.inspect}, #{o.inspect})”);  
end  
end

irb(main):096:0\> YourClass.new + 1  
=\> #\<YourClass:0x1016bbd0 @x=“addition(#\<YourClass:0x1016bc78 @x="foo"\>,  
1)”\>  
irb(main):097:0\> 1 + YourClass.new  
=\> #\<YourClass:0x10164940 @x=“addition(#\<YourClass:0x10164b98 @x=1\>,  
#\<YourClass:0x10164ce8 @x="foo"\>)”\>  
irb(main):098:0\> YourClass.new - 1  
=\> #\<YourClass:0x100cda68 @x=“substraction(#\<YourClass:0x100cdb58  
@x="foo"\>, 1)”\>  
irb(main):099:0\> 1 - YourClass.new  
=\> #\<YourClass:0x100c2a28 @x=“substraction(#\<YourClass:0x100c2ad0 @x=1\>,  
#\<YourClass:0x100c2b48 @x="foo"\>)”\>

Note: coerce might have to become more complex depending on the argument  
types. Typically you do something like:

def coerce(o)  
case o  
when Fixnum, Bignum  
…  
when Float  
…  
when String  
…  
when SomeOtherType  
…  
else  
raise ArgumentError, “Cannot coerce #{o.inspect}”  
end  
end

Regards

```
robert

```

---

<div class="post-metadata">

**Author:** ![Mark10](https://avatars.discourse-cdn.com/v4/letter/m/df705f/32.png) [@Mark10](https://rubytalk.org/u/Mark10)\
**Post date:** [29 April 2004 19:45 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/5 "2004-04-29T19:45:49Z")

</div>

Jean-Hugues ROBERT wrote:

> > 1. Can I redefine coerce() ? If so, when is it called ? (I tried,  
> > it is not called…)
> > 
> > 2. By defining to\_str() I can have “xx” + MyClass.new(xx) work. What  
> > do I have to redefine to have 1 + MyClass.new(xx) work ?

this works for me

class MyClass  
def initialize(val)  
@value = val  
end

```
def coerce(other)
   [other, @value]
end

```

end

1 + MyClass.new(5)  
#=\> 6

> > Google “ruby coerce” provided little help.

You can find some information by searching the ruby-talk archives  
[http://blade.nagaokaut.ac.jp/ruby/ruby-talk/index.shtml](http://blade.nagaokaut.ac.jp/ruby/ruby-talk/index.shtml) for coerce.

HTH

> **···**
>
> > At 03:45 30/04/2004 +0900, you wrote:
> 
> –  
> Mark Sparshatt

---

<div class="post-metadata">

**Author:** ![Jean-Hugues\_ROBERT](https://avatars.discourse-cdn.com/v4/letter/j/e47c2d/32.png) [@Jean-Hugues\_ROBERT](https://rubytalk.org/u/Jean-Hugues_ROBERT)\
**Post date:** [30 April 2004 05:04 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/6 "2004-04-30T05:04:38Z")

</div>

Thanks a lot.

Looking for some example code I had had a look at  
ruby/lib/ruby but there were so many files that I was  
slightly lost.

Is there something like  
Class.implementers( meth)  
that would return a list of all known classes that implement  
some method ? It could not returned the classes that where  
never required but it may help understand the protocols for  
common methods that where designed to be redefined

Additionally, is there something like  
a\_class.sources()  
that would return an array of “positions” where source code  
for a a\_class was “loaded”. It would return something similar  
to what you get when an exception bombs at these positions.

If so, then  
Class.implementers( :coerce).each do | klass |  
p “coerce() is implemented by #{klass}, maybe in:”  
p a\_class.sources.join( “\n”)  
end  
would list the files where there might be examples of how  
to implement the :coerce method.

If I come up with a solution for this issue with Introspection,  
I will share it here.

Thanks in advance,

BTW: With all the help I received I am well on my way to implement  
a Reference class, like a Pointer but with some auto-dereferencing.  
OTOH maybe I should investigate a “proxy” type of solution, like  
the ones used by Drb or similar stuff.

Yours,

Jean-Hugues

> **···**
>
> At 08:39 30/04/2004 +0900, you wrote:
> 
> > Jean-Hugues ROBERT wrote:  
> > …
> > 
> > > Google “ruby coerce” provided little help.  
> > > Thanks in advance,
> > 
> > The coerce framework makes up for the lack of  
> > multiple dispatch aka “method overloading”,  
> > whose lack can be a major nuisance when doing  
> > numerical work. You probably should study the  
> > source code of the standard libraries “complex”  
> > or "rational, if you want to get a feel for  
> > using this framework. Here is a scaled down  
> > example
> > 
> > * * *
> > 
> > class MyClass
> > 
> > def initialize(val)  
> > @value = val  
> > end
> > 
> > def coerce(other)  
> > [MyClass.new(other),self]  
> > end
> > 
> > def -(lhs)  
> > if MyClass === lhs  
> > MyClass::new(@value - lhs.value)  
> > else  
> > x,y = lhs.coerce(@value)  
> > MyClass::new(x - y)  
> > end  
> > end
> > 
> > protected  
> > attr\_reader :value  
> > end
> > 
> > p 1 - MyClass.new(5)  
> > p MyClass.new(5) - 1
> > 
> > require ‘complex’
> > 
> > z = Complex(1.0,1)
> > 
> > ## p z - MyClass.new(5) p MyClass.new(5) - z
> > 
> > ## #\<MyClass:0x28a63e8 @value=-4\> #\<MyClass:0x28a6358 @value=4\> #\<MyClass:0x28955b8 @value=Complex(-4.0, 1)\> #\<MyClass:0x28954e0 @value=Complex(4.0, -1)\>
> > 
> > /Christoph
> 
> * * *
> 
> Web: [@jhr is virteal, virtually real](http://hdl.handle.net/1030.37/1.1)  
> Phone: +33 (0) 4 92 27 74 17

---

<div class="post-metadata">

**Author:** ![Jean-Hugues\_ROBERT](https://avatars.discourse-cdn.com/v4/letter/j/e47c2d/32.png) [@Jean-Hugues\_ROBERT](https://rubytalk.org/u/Jean-Hugues_ROBERT)\
**Post date:** [30 April 2004 05:04 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/7 "2004-04-30T05:04:40Z")

</div>

> Jean-Hugues ROBERT wrote:
> 
> > > 1. Can I redefine coerce() ? If so, when is it called ? (I tried,  
> > > it is not called…)
> 
> class MyClass  
> def initialize(val)  
> @value = val  
> end
> 
> def coerce(other)  
> [other, @value]  
> end  
> end
> 
> 1 + MyClass.new(5)  
> #=\> 6

For me too. I must have been sleeping when I tried it.  
Thanks !

> > > Google “ruby coerce” provided little help.  
> > > You can find some information by searching the ruby-talk archives  
> > > [http://blade.nagaokaut.ac.jp/ruby/ruby-talk/index.shtml](http://blade.nagaokaut.ac.jp/ruby/ruby-talk/index.shtml) for coerce.

That link does not work for me ☹

I tried with Search (Namazu): coerce =\>  
Namazu: a Full-Text Search Engine  
This index contains 0 documents and 0 keywords.

I tried with (Recent 1000 files, Regular Expression): coerce =\>  
TIMEOUT may occur. Using Netscape is recommended.

I tried (Subjects, Regular Expression): coerce =\>  
TIMEOUT may occur. Using Netscape is recommended.

It had happened to me before.  
I must be out of luck today too 😉

Thanks anyway.

Jean-Hugues

> **···**
>
> At 04:45 30/04/2004 +0900, you wrote:
> 
> > > At 03:45 30/04/2004 +0900, you wrote:
> 
> > HTH
> > 
> > –  
> > Mark Sparshatt
> 
> * * *
> 
> Web: [http://hdl.handle.net/1030.37/1.1](http://hdl.handle.net/1030.37/1.1)  
> Phone: +33 (0) 4 92 27 74 17

---

<div class="post-metadata">

**Author:** ![Harry\_Ohlsen1](https://avatars.discourse-cdn.com/v4/letter/h/a88e57/32.png) [@Harry\_Ohlsen1](https://rubytalk.org/u/Harry_Ohlsen1)\
**Post date:** [30 April 2004 05:14 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/8 "2004-04-30T05:14:32Z")

</div>

> Is there something like  
> Class.implementers( meth)  
> that would return a list of all known classes that implement  
> some method ? It could not returned the classes that where  
> never required but it may help understand the protocols for  
> common methods that where designed to be redefined

If you have ri installed (and if you don’t, I’d definitely recommend doing so), try something like

ri

For example, “ri open” gives me

The method named `open’ is not unique among Ruby’s classes and modules:  
Dir::open, File::open, Kernel::open

Cheers,

Harry O.

---

<div class="post-metadata">

**Author:** ![Laurent\_Julliard4](https://avatars.discourse-cdn.com/v4/letter/l/ebca7d/32.png) [@Laurent\_Julliard4](https://rubytalk.org/u/Laurent_Julliard4)\
**Post date:** [30 April 2004 08:54 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/9 "2004-04-30T08:54:01Z")

</div>

this?

ObjectSpace.each\_object(Class){|x| p x if  
x.instance\_methods.member?‘coerce’}

> **···**
>
> il Fri, 30 Apr 2004 14:04:38 +0900, Jean-Hugues ROBERT [jean\_hugues\_robert@yahoo.com](mailto:jean_hugues_robert@yahoo.com) ha scritto::
> 
> > Is there something like  
> > Class.implementers( meth)  
> > that would return a list of all known classes that implement  
> > some method ?

---

<div class="post-metadata">

**Author:** ![Mark10](https://avatars.discourse-cdn.com/v4/letter/m/df705f/32.png) [@Mark10](https://rubytalk.org/u/Mark10)\
**Post date:** [30 April 2004 12:07 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/10 "2004-04-30T12:07:06Z")

</div>

Jean-Hugues ROBERT wrote:

> That link does not work for me ☹
> 
> I tried with Search (Namazu): coerce =\>  
> Namazu: a Full-Text Search Engine  
> This index contains 0 documents and 0 keywords.

That’s a pecularity of the search engine. If you scroll down the page  
you’ll see the list of results (there are 209 for coerce)

> **···**
>
> > I tried with (Recent 1000 files, Regular Expression): coerce =\>  
> > TIMEOUT may occur. Using Netscape is recommended.
> > 
> > I tried (Subjects, Regular Expression): coerce =\>  
> > TIMEOUT may occur. Using Netscape is recommended.
> > 
> > It had happened to me before.  
> > I must be out of luck today too 😉
> > 
> > Thanks anyway.
> > 
> > Jean-Hugues

---

<div class="post-metadata">

**Author:** ![Jean-Hugues\_ROBERT](https://avatars.discourse-cdn.com/v4/letter/j/e47c2d/32.png) [@Jean-Hugues\_ROBERT](https://rubytalk.org/u/Jean-Hugues_ROBERT)\
**Post date:** [30 April 2004 16:12 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/11 "2004-04-30T16:12:05Z")

</div>

I tried.  
Apparently something is broken.  
It just complains about a bad path.

I had a look at ri.bat and apparently it references some weird  
E:\Dev\RubyDev\rubyinstaller.…  
I tried to fix the offending line. But then I get a msg  
about missing ri documentation and some rdoc run to create  
documentation.

I ran “rdoc” but it parsed only the files from the directory where  
I invoked it (c:\ruby\bin in that case). So I ran it again, but  
from c:\ruby this time. That took a while…

Then rdoc what about generate HTML but  
“Directory doc already exists, but it looks like it isnt’…”  
and rdoc asked me to re-run using some --op option.

I suppose this is when I gave up. What should have I done ?

This might be the Windows packaging issue mentioned in some msg  
threads a while ago, I haven’t read the msgs carefully.

Thanks for the suggestion anyways. Investigating “ri” is on my  
“to do” list now.

Yours,

Jean-Hugues

> **···**
>
> At 14:14 30/04/2004 +0900, you wrote:
> 
> > > Is there something like  
> > > Class.implementers( meth)  
> > > that would return a list of all known classes that implement  
> > > some method ? It could not returned the classes that where  
> > > never required but it may help understand the protocols for  
> > > common methods that where designed to be redefined
> > 
> > If you have ri installed (and if you don’t, I’d definitely recommend doing  
> > so), try something like
> > 
> > ri
> > 
> > For example, “ri open” gives me
> > 
> > The method named `open’ is not unique among Ruby’s classes and modules:  
> > Dir::open, File::open, Kernel::open
> > 
> > Cheers,
> > 
> > Harry O.
> 
> * * *
> 
> Web: [@jhr is virteal, virtually real](http://hdl.handle.net/1030.37/1.1)  
> Phone: +33 (0) 4 92 27 74 17

---

<div class="post-metadata">

**Author:** ![Jean-Hugues\_ROBERT](https://avatars.discourse-cdn.com/v4/letter/j/e47c2d/32.png) [@Jean-Hugues\_ROBERT](https://rubytalk.org/u/Jean-Hugues_ROBERT)\
**Post date:** [30 April 2004 16:12 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/12 "2004-04-30T16:12:11Z")

</div>

I haven’t played with ObjectSpace yet. Now I can see that it is  
an interesting thing ! Thanks.

What I need is almost what you propose. The difference is that  
I need to know what classes do actually implement some method,  
versus inherit an existing implementation.

i.e.  
class X; def meth() end end  
class Y \< X; end  
Class.implementers( :meth) # =\> X does implement :meth, Y merely inherit it.

It is easy to assert that X actually implements :meth, because  
X.superclass does not. However I still don’t know how to  
filter out Y.

Next step will be to figure out “where” X defines :meth.  
A refactoring editor could use that to point to :meth  
implementations that may need some refactoring.

Or, regarding my initial need, I could know which file to look  
at where :coerce is defined (so that I can understand how it  
works, which is something I don’t need to do anymore, thanks  
to all the gentle answers to my initial question).

Yours,

Jean-Hugues

> **···**
>
> At 17:54 30/04/2004 +0900, you wrote:
> 
> > il Fri, 30 Apr 2004 14:04:38 +0900, Jean-Hugues ROBERT \>[jean\_hugues\_robert@yahoo.com](mailto:jean_hugues_robert@yahoo.com) ha scritto::
> > 
> > > Is there something like  
> > > Class.implementers( meth)  
> > > that would return a list of all known classes that implement  
> > > some method ?
> > 
> > this?
> > 
> > ObjectSpace.each\_object(Class){|x| p x if  
> > x.instance\_methods.member?‘coerce’}
> 
> * * *
> 
> Web: [@jhr is virteal, virtually real](http://hdl.handle.net/1030.37/1.1)  
> Phone: +33 (0) 4 92 27 74 17

---

<div class="post-metadata">

**Author:** ![Laurent\_Julliard4](https://avatars.discourse-cdn.com/v4/letter/l/ebca7d/32.png) [@Laurent\_Julliard4](https://rubytalk.org/u/Laurent_Julliard4)\
**Post date:** [30 April 2004 18:44 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/13 "2004-04-30T18:44:00Z")

</div>

that was the broken ri pkg. see a thread appeared here later, try the  
newer installer, or look at [ruby-doc.org](http://ruby-doc.org) for link on where to download  
ri data files and how to setup them

> **···**
>
> il Sat, 1 May 2004 01:12:05 +0900, Jean-Hugues ROBERT [jean\_hugues\_robert@yahoo.com](mailto:jean_hugues_robert@yahoo.com) ha scritto::
> 
> > I tried.  
> > Apparently something is broken.  
> > It just complains about a bad path.
> > 
> > I had a look at ri.bat and apparently it references some weird  
> > E:\Dev\RubyDev\rubyinstaller.…

---

<div class="post-metadata">

**Author:** ![Jean-Hugues\_ROBERT](https://avatars.discourse-cdn.com/v4/letter/j/e47c2d/32.png) [@Jean-Hugues\_ROBERT](https://rubytalk.org/u/Jean-Hugues_ROBERT)\
**Post date:** [30 April 2004 20:38 UTC](https://rubytalk.org/t/coerce-what-protocol-to-implement-it/11291/14 "2004-04-30T20:38:01Z")

</div>

Will do. Thanks. EOM

> **···**
>
> At 03:44 01/05/2004 +0900, you wrote:
> 
> > il Sat, 1 May 2004 01:12:05 +0900, Jean-Hugues ROBERT \>[jean\_hugues\_robert@yahoo.com](mailto:jean_hugues_robert@yahoo.com) ha scritto::
> > 
> > > I tried.  
> > > Apparently something is broken.  
> > > It just complains about a bad path.
> > > 
> > > I had a look at ri.bat and apparently it references some weird  
> > > E:\Dev\RubyDev\rubyinstaller.…
> > 
> > that was the broken ri pkg. see a thread appeared here later, try the  
> > newer installer, or look at [ruby-doc.org](http://ruby-doc.org) for link on where to download  
> > ri data files and how to setup them
> 
> * * *
> 
> Web: [http://hdl.handle.net/1030.37/1.1](http://hdl.handle.net/1030.37/1.1)  
> Phone: +33 (0) 4 92 27 74 17