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

Re: [zzdev] Re: [zzdev] Notes on accessing past versions



On 20010223T224802+0100, Benjamin Fallenstein wrote:
> As you probably noticed, I committed an alternative which I think is
> "right:" version:d.1 shows the connection in whatever version the given
> cell is in.

Yes, that's better, thanks.  But IMHO the "right" solution would be to
map old dimension d.1 into current dimension d.1.  IF it can be done
without calling up all sorts of invariant violations.

> >   - The version:list dimension is deep magic and in some ways (namely, in how
> >     it handles connecting along it) violates the ZigZag paradigm.  Tell me what
> >     you think about it.
> 
> Okay, if you want to. ;o) I think it's all wrong.

Okay.

> I think a version
> should be activated by accessing some cell in that version-- and that
> works, I've tried

Yes, it works.  (It was designed to work.  That was the simplest way
to make activations semipersistent.)

> Now, why bother the user about "active"
> and "passive" versions?

Because activation is a potentially slow operation.  And active versions
are potential memory hogs.  The user might want to explicitly deactivate
a version.

> The version:list dimension should just be all
> natural numbers from 1 to <current timestamp>. Rotating from
> version:list to version:home would activate the specific version used.

No, that would activate all versions that the view happened to traverse
(at least all shown on screen).

-- 
%%% Antti-Juhani Kaijanaho % gaia@xxxxxx % http://www.iki.fi/gaia/ %%%