Re: [bug] Inheritance and friend access control broken

[Robert Lipe <robertl at dgii dot com>]

> > So, the experts have checked appropriate standards and agree this is 
> > a problem, right?
> 
> Yup.

Lovely.  That's the part I can't help much with.

> > Any volunteers to create a testcase?
> Here it is.

I've now committed this.   Thank you.

> BTW, is there any easy way to create a patch file that will add a new
> file to the CVS tree, if you don't have write access to it? `cvs diff
> -N friend1.C' won't do it, because CVS doesn't know that file :-(

I don't know.  I'll confess that I stuck the file in and did a 'cvs add'
on it before I tried the 'cvs diff -N' which then did work, of course.
Perhaps a diff -N outside of CVS against /dev/null or something?  

> When I submit patches, should my changelog entries contain a ChangeLog 
> header (timestamp + name + e-mail address) or not?

I pasted them out of your header.  ChangeLog convention has it that the
timestamp is the time of the commit, not the time of the submission.
It's going to be more accurate (at least for me) to smoosh in a new date
than for me to misspell both of your names recreating the header. :-)


-- 
Robert Lipe       http://www.dgii.com/people/robertl       robertl@dgii.com