[ticker-dev] Re: channel subscription rules

[Ian Lister]

On Fri, 23 Apr 2004, David Arnold wrote:
>I think it's more a case of preferring to distribute the work, rather
>than centralise it.

That seems a little antithetical to the Elvin approach of pushing the work
into the network rather than leaving the edge nodes to do it :-)

>But .. it's not a great deal of work, and it's something that's pretty
>simple for the Elvin Router to do, so it's not a major consideration.

But do you still think it's more important than the ease of use gained?

>  Ian> While Sticker may be a model citizen in implementing many of
>  Ian> these new protocol features (they seem to have a tendency to go
>  Ian> from Sticker to the protocol rather than the other way around)
>
>Which I think is an excellent model, FWIW.

Indeed. I guess I didn't consider that `model citizen' might be taken to
mean anything but good :-)

>So, what's the general opinion?  Is everyone happy to move to using
>
>  decompose-compat(fold-case(Group)) == decompose-compat(fold-case("MyGroup"))
>
>as the general subscription template ?

I am.

It needs to be accompanied by rules about copying the value exactly when
replying, though.

Ian