On 6/17/08, Brian Ford <brixen / gmail.com> wrote:
> Hi all,
>
>
>  On Jun 16, 8:00 pm, Urabe Shyouhei <shyou... / ruby-lang.org> wrote:
>  > Vladimir Sizikov wrote:
>  > > I"m trying to understand why it was needed, and was it really needed
>  > > that bad to add to 1.8.6 series?
>  >
>
> > It was introduced to fix ruby-bugs #19377http://rubyforge.org/tracker/index.php?func=detail&group_id=426&atid=...
>
>  My 2 cents on this: The #respond_to? method allows you (the
>  programmer) to optionally search for private methods (which you can
>  then call with #send). But the decision to search for private methods
>  should always be in the hands of the programmer. The language and libs
>  should never make this decision for you. If I choose to make methods
>  "private" in my code, I expect other code to respect that if it wants
>  to play nicely. I *never* expect the system to ignore my intentions in
>  some ad hoc manner.
>
>  Also, as a general principle, a bug fix should touch only the code and
>  functionality centered around the bug. This change affects code and
>  functionality that is very distant from the problem.
>
>  I am strongly in favor of reverting this change and using a fix that
>  only focuses on Marshall. If #_load and #_dump need to be public
>  instead, that would make a lot more sense to me.
>

I agree. I don't understand why _load and _dump are private.
They are almost always called from somewhere other than self, are they not?
Was the intent simply to hide them from being output in 'methods' and
'instance_methods' lists?