Thanks for the bug report, how to reproduce this?
Particularly: what architecture, loaded libraries?
Hm, I truly don't know. I have a Gentoo system with ruby-1.8.2.
I just type "env -i ruby -rihelp -e 'Array.help'" at the prompt, and
this happens. "ri Array" works fine. Neither is a problem with the
ruby-cvs ebuild.
Hm, I looked around it, I have a
/usr/share/ri/1.8/system/TSortArray/cdesc-TSortArray.yaml
file which appears to be something that shouldn't be there/ I don't know
how is this generated (I don't have this file either in my 1.9
installations or in an other 1.8.2 installation). I wonder is there
anybody else with such a file around?
It's strange though, Array.help really shouldn't go _down_
in the inheritance hierarchy. First time for everything I suppose.
Apart from that flaw in my installation, this is why it still can
inidcate a bug in IHelp: why does it give priority to this stub subclass
documentation? Add to this that ri doesn't do that...
Here is the content of cdesc-TSortArray.yaml if you want to test the
behaviour of IHelp in the presence of that file:
--- !ruby/object:RI::ClassDescription
attributes: []
class_methods: []
comment:
constants: []
full_name: TSortArray
includes: []
instance_methods: []
name: TSortArray
superclass: Array
>* Extension:
>
>ihelp-dump.rb, you find it at http://www.creo.hu/~csaba/ruby/ with
>generated rdoc.
>
>It enables you to dump rendered docs to any object which features #<<.
(snip)
Thank you for this aswell, it looked like it might come handy on a
quick glance.
Well, in
http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/127842
you told me that by your knowledge piping in the output of ri is the
only way to get the rendered text, so I had to find a better way of
doing that, of course 
Csaba
···
On Fri, Mar 04, 2005 at 10:52:15AM +0900, Ilmari Heikkinen wrote: