[PATCH] Yet another bug in NormalExpression's restrict()

[Nick Hilliard]

On 06/07/2009 03:55, Faidon Liambotis wrote:
> restrict() in NormalExpression seems not to have been thought out well.
> It didn't work properly in anything but simple v4/v6 prefixes, probably
> needs some serious rework.
>
> This time, the bug is that it set the singleton_flag to PRFX/IPV6_PRFX
> if it found a non-empty prefix list inside the term. It didn't account,
> however, for terms containing a prefix list *and* something else (e.g.
> a community) or for expressions containing multiple, different, terms.

hmmm, looks ugly alright.  Can you create a bug in trac to cover this just 
so that the problem doesn't get lost or masked by this fix?  Test cases 
displaying the brokenness would be really good. - thanks!

r290.

Nick