[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]

Re: [zzdev] d.cellcreation



On 13 Feb, Antti-Juhani Kaijanaho wrote:
> Anyway, I noticed that d.cellcreation no longer traverses the whole
> space. Its current semantics is: list all cells which have a natural
> number as their ID. However, there are cells nowadays that don't have
> a simple natural number as their ID: foreign cells (cells created
> originally in another space and locally stored here for convenience,
> for example after ZTP GET) and spacepart cells, for example.

d.cellcreation was first just an example of an implicit dimension. I
remember the implementation was planned to be replaced by something more
advanced as soon as the IDs wouldn't be just natural numbers. So the
semantics are better put "list all cells in the space in creation
order". I'd only include the permanent cells, though, because that's
what the dimension is good for: getting a list of all user-made and
other "data" cells. Getting "data" for a space part shouldn't be done by
taking all the cells (there can be infinite number of them), but by
taking the permanent cells which describe that space part.

> Do we change the d.cellcreation semantics to include foreign cells
> too? If we do, do we want any ordering properties to
> d.cellcreation?

I don't know what the situation with slices is. However, in addition to
d.cellcreation, d.slices would get traversed when finding permanent
cells. As I see it, it would be natural to create one slice per foreign
space, and keep the foreign cells in their respective slices, ordered
(by creation time) along d.cellcreation. Could it be the server's task
to send also d.cellcreation connections?


The way I see it,
Tuukka

--
-- Trying to catch me? Just follow up my Electric Fingerprints
-- To help you: Tuukka.Hastrup@xxxxxx
                http://www.iki.fi/Tuukka.Hastrup/
                ICQ #11321669