[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]
Subspaces & ZZ versioning
- To: Antti-Juhani Kaijanaho <gaia@xxxxxx>
- Subject: Subspaces & ZZ versioning
- From: Benjamin Fallenstein <b.fallenstein@xxxxxx>
- Date: Thu, 21 Dec 2000 20:01:39 +0100
- Cc: zzdev@xxxxxxxxxx
- References: <20001210185537.D15044@xxxxxxxxxxxxxx> <3A33DE1D.375C90A6@xxxxxx> <20001210224318.A17858@xxxxxxxxxxxxxx> <3A34009B.FB990E0@xxxxxx> <20001211015201.E18012@xxxxxxxxxxxxxx> <3A342176.B03A9B9F@xxxxxx> <20001211035226.B3587@xxxxxxxxxxxxxx> <3A3E67F4.3C00B087@xxxxxx> <20001219124905.A7717@xxxxxxxxxxxxxx> <3A41351B.BA16A23D@xxxxxx> <20001221111527.B29066@xxxxxxxxxxx>
Hi A-J!
Antti-Juhani Kaijanaho wrote:
> On 20001220T233923+0100, Benjamin Fallenstein wrote:
> > Yes, I'd prefer xml too, but the xml classes needed are unhandy and pose
> > too many problems. The Dump format, along with subspace selectors (which
> > need to wait some time pbly), should make it possible to use zzspaces in
> > CVS. (The reverse section is there to make diffs usable.)
>
> What's your image of subspace selectors? Do you have something written
> up?
No, sorry. Actually, I'm currently thinking of something really
primitive to get the thing going-- i.e., when you edit the subspace,
it's connected to the main space only through a special dimension, so
you start at a cell and follow all dimensions for the subspace (probably
its best to give a list), and all cells you reach are included.
That, of course, is only a start. Actually, I think in the long run we
need a general interface where you can plug in arbitrary algorithms
which select a number of cells and connections.
Why? Because we need applitudes to be able to provide tools which select
subspaces as entities which "make sense" (Nile selections; calendar
weeks; function definitions; etc.). That's important for versioning, for
creating distributable bunches, etc.
I predict most selectors will follow the pattern: Select a number of
cells in some special way. Select all connections between these cells on
a limited set of dimensions. (We should, though, also allow selecting
only certain connections.)
> I'm interested because I'm designing a ZZ network protocol for
> an eventual CVS-like distribution and version control system for GZZ,
> and I have my own sketch, and I'd like them to be at least compatible.
I'm really interested in that. Do you have time to explain a bit? I'd
love to talk about it. (Might also be able to do a bit of coding, if
that'd help you.)
Which kind of version control do you have in mind? Do you remember my
earlier notes about pull-outs? They're something I consider very
important and, should I have time, would actually like to try -- thing
is, I probably won't have time. :(
Generally, I'd like a versioning system for GZZ much better than the
dump thing, but I do want to start some serious coding with my
"visual"/"natural language" programming applitude, and there are just
too many interconnections to write the necessary standard space in Java.
I suppose I'll make Dump very simple and stop using it the very moment
your system's ready. ;o)
- Benja