[ANN] BabySRE (Squeak Reverse Engineering) posted on SqueakMap
Ned Konz
ned at squeakland.org
Thu Dec 23 03:59:13 UTC 2004
On Wednesday 22 December 2004 12:36 am, hjh-sqlist at lexdb.net wrote:
> I suppose that a lot of additional expertise gained through the
> process went into Connectors2.0 and I will read in the code, when I have
> time . Are there any things to say about API changes?
From the point of view of simple use (creation, subclassing, etc.), the API is
about the same. As Stephane Rollandin points
I have added hooks for connecting and maintaining a graph model that parallels
the structure of the connectors and morphs in your drawing. This is the
biggest change to the API in v2. I originally had a version that included the
model classes themselves (with a property dictionary for end-user extension).
However, as I worked on it I removed all of the graph model stuff except for
the hooks. The problem was that the uses I have and see for Connectors have
different models of what a hierarchical graph is. I couldn't figure out a way
to model the differing views of hierarchical graphs without getting in the
way of later extension and use.
I've been reading and trying to digest a PhD thesis paper by Giorgio Busatto
that is a kind of "grand unified theory" of hierarchical graphs. He describes
a considerably more intricate structure that can represent all the different
kinds of hierarchical graphs that he was aware of by reading other work and
using other software. I haven't done any work with it besides reading and
taking notes, though, as I've been busy with other things.
> On the other side from a user point of view SRE is just fine as is.
>
> In any case, thank you Ned for this wonderful contribution of Connectors,
> with which you put others into a position to develop exciting things.
> (Jacaranda, SRE ...)
You're welcome!
--
Ned Konz
http://bike-nomad.com/squeak/
More information about the Squeak-dev
mailing list
|