On 2013-09-18, at 05:54, Steve Thomas sthomas1@gosargon.com wrote:
On Tue, Sep 17, 2013 at 2:36 PM, karl ramberg karlramberg@gmail.com wrote: This is IMO a bug. It says tellAllSuccessors but tells only the first node in the successor graph.
Agreed, though the bug is only in the naming, the behavior is correct. Mathematically speaking we mean the direct successors, but for brevity this got shortened to just successors.
Note that the help bubble for "tell all outgoing connections" is very carefully worded:
"Send a message to all the connectors whose source end is connected to me"
whereas the help for "tell all successors" is just
"Send a message to all graph successors"
which is both factually wrong (it's only direct successors) and should include a mention of connectors, perhaps
"Send a message to all objects at the destination end of my outgoing connections"
Well looking at this again, it seems using "tell all" is a bad wording choice. All of the "tell" tiles are "tell all" tiles. Perhaps it would be better to have two versions "tell" tiles (for connectors and successors) which tell the 1st level and "tell all" which go all the way down/up the chain.
All of the current behavior in Etoys is centered on a single object: *my* siblings, *my* children, *my* connections. This comes from the object-centric view of Etoys. So "all" means "all the objects I know" not "all objects there are" which would be a global view, not object-centric. I'm not convinced that we should give up that very fundamental property of Etoys for a little convenience in a very limited scenario.
And also get rid of the extraneous "all" in the other tell commands. "tell all siblings" ==> "tell siblings" "tell all embedded objects" ==> "tell embedded objects" "tell all incoming connections" ==> "tell incoming connections" and "tell all incoming connections" Of course the problem with #3 and its like is that it could break older projects which would not want to do. You might be able to determine which version it was saved in and deal with that I don't know.
I like the "all" in the wording because it alerts the user that something is happening to multiple objects at once. It is not strictly necessary because with careful reading the plural "s" will reveal as much. I still think a single letter is not quite enough to make clear that this is, in fact, a very unusual thing, compared to the normal explicit object relationships.
- Bert -