basic vs full image
edw at generalmagic.at
Tue Dec 14 10:17:40 UTC 2004
when I do smalltalk development (I do mostly business apps,
but I also work on software generation technology),
I use a capable development machine, and try to load "everything".
Even when developing product families (fat client/client server/server
only), I try to structure the code so that all the stuff can
be loaded simultaneously, and the product flavor to run is
This way, when I do refactorings, I will always find *all* references,
*all* implementors etc, and can fix those while I am at it.
I believe this is advantegous because
1. I know what I am doing when I do the refactoring, it is easy to
change all references compared to later when something breaks
(or does not even load).
2. It is not "somebody elses problem" when somebody elses code breaks
because of my refactorings; I dont want to piss people off.
3. Things do not add to my (or somebody elses) to-do list; when I'm
done, I'm really done (I hope this means what I think it means; I am
not a native speaker).
4. The refactorings are better when you have the "full picture"
5. when looking for examples, the likelihood that the code is already
there and does not need to be searched, and loaded, is higher,
meaning better productivity.
In a distributed development, this would imply that I would sometimes
touch a lot of code that is not "mine". Therefore a tool is needed
to be able to distribute these changes back in a configuration managed
(I use envy (you probably guessed already)).
When I look at Squeak, the pleasant experience (for me) is how much
is there, how rich the environment is, how much things can be done.
For me, the full image is also the most appropriate for programming,
and the more there is in it, the better.
When most people would use the full image, the things would work more
stable (because they have been tried more often), and the whole codebase
can be advanced in a more consistent way.
Just my opinion, of course...
More information about the Squeak-dev