ruby-1.8.1p4 and ri

[Fritz Heinrichmeyer]

Hello,

maybe this is only a freebsd question:

i had a little frustration with ri18 as found in the newest tarball as
there are problems in some files when i try to rebuild the "ri" database
(yaml/stringio.rb i.e.) using rdoc. "rdoc" is a little unforgiving and
does not produce anything if it finds something it regards as error. I
had to convince rdoc with xargs.

I have to install the port ruby-ri apart from it to get usefull ri
documentation. The port produces a usefull script called ri ...

Any pointer to the recommended way of rebuilding ri documentation from
source is appreciated.

 

[Joey Gibson]

Hello,

maybe this is only a freebsd question:

Nope. Win2k as well.

i had a little frustration with ri18 as found in the newest tarball as
there are problems in some files when i try to rebuild the "ri" database
(yaml/stringio.rb i.e.) using rdoc. "rdoc" is a little unforgiving and
does not produce anything if it finds something it regards as error. I
had to convince rdoc with xargs.

I saw the same problem on Win2k. The error was because the stringio file
was referring to the YAML module, but it did not declare itself as part
of the YAML module, nor did it reference it. I got rdoc to run properly
by buttressing everything in the stringio file with

module YAML
...
end

Whether that's correct or not I don't know.

--
Never trust a girl with your mother's cow,
never let your trousers go falling down in the green grass...

http://www.joeygibson.com
http://www.joeygibson.com/blog/life/Wisdom.html

 

[Dave Thomas]

Any pointer to the recommended way of rebuilding ri documentation from
source is appreciated.

I'm working slowly through stuff to get this resolved. In this
particular case, the source file is question is a tad messy, and it's a
StringIO replacement that ends up modifying the a method back in the
YAML module. Because the YAML module isn'y defined at this point, RDoc
didn't know what to do.

For now, I've changed RDoc to assume that any unknown constant that
prefixes a method name is a module: we'll see how that works out in
practice. Longer term, I think I'll refactor the yaml code to use a
factory and get rid of this issue at the source level.

I'll check this change in later today.


Cheers

Dave