Re: pipapo with element shadowing, wildcard support

[Florian Westphal <fw@strlen.de>]

Stefano Brivio <sbrivio@redhat.com> wrote:
> > Question would be if it makes sense to relax the -EEXIST checks so that in
> > step 1 the wider key could be inserted first and then allow it to be
> > (partially) shadowed later.
> 
> The reason why I implemented it this way was to avoid possible
> ambiguity because entries inserted first are anyway matched first.
> Details with example at:
> 
>   https://lore.kernel.org/all/20200226115924.461f2029@redhat.com/

Thanks.

> It might make sense to change that, but that's not entirely trivial as
> you would need to renumber / reorder entries in the buckets on
> insertion, so that more specific entries are always added first.

Right, it needs a reorder step.

> I guess you assumed this was already implemented, which is a reasonable
> assumption, but unfortunately I didn't add that, it looked good enough
> as it was, back then.

No, I realized this was missing, the -EEXIST tests would not have
made sense otherwise.

What I did not know if it was omitted due to 'not feasible/too hard' or
'good enough for now'.

> See that same thread for some discussion about it, in particular around:
> 
>   https://lore.kernel.org/all/20200225184857.GC9532@orbyte.nwl.cc/
> 
> other than that I'm not aware of previous discussions.

Thanks for the pointer!