Two important issues...
danielv at netvision.net.il
Sat Feb 15 13:20:19 UTC 2003
Roel Wuyts <roel.wuyts at iam.unibe.ch> wrote:
> I agree with your comments IF we would want to release Traits as a
> 'product', or if we would be paid somehow to do this... But
> (1) we don't have the resources to do this (we have to keep moving -
> Traits TNG and new modules are coming), and
I know you are not under any contract to deliver perfect product and
documentation to me, your salary gets paid for doing research (Wish I
was so lucky). And I never said you have to do anything. I didn't even
imply you *should* do something. I suggested you might want to keep some
things in mind, if you want people to use your code (as Nathanael said
he does). Whether you do that because it contributes to your research to
publish what other people have done with the model, or because of the
fuzzy warm feeling it gives you is up to you.
I mentioned the things I did simply because in my experience, they help
get things adopted. Please don't read this thread as being about what I
"expect" of you, or whether I "appreciate" your work. Hey, I got a
personal demo from Nathanael, and I was and remain very impressed. My
opinion is that it's a sound model, and can have great benefits for
There's a distance between a model and a product, and even many
wonderful products don't get used. But if you want it to be used, you DO
have to think about it a little like it's a product.
> (2) some work Nathanael did was lost already (the ClassBuilder fix,
> specifically, which is an amazingly difficult fix to do...). Very, very
I'm sorry, my best understanding of that thread was that the
ClassBuilder fix that was done simply fixed a different bug. If we as
harvesters missed a big one, please let us know, we have an alpha cycle
coming up... but that's a different thread.
> Currently the Traits work is in a pretty reasonable state to play with.
> However, on the list I only hear people that doubt it will work or that
> want more info. But there is code! Really, just download the image and
> play with it. Don't take our word or a demo of some older version you
> saw. Treat it as the other things that come bundled with Squeak and
> that are cutting edge.
Hey, I don't doubt that it works, I said something very specific and
simple. The screenshots don't show me how to do anything with the tool,
they just show where the buttons are. Your group has mentioned that
you've used the tool to do refactorings. LOTS of people are drooling
about the results you've mentioned, but you don't show how to reproduce
them. It would really help if you did. It's really pretty silly that we
all now start to guess "what does this button do" when what you really
want to present is a novel model of code, and a novel way of refactoring
> Was there ever a decent description of Morphic
> when it came out? No. And it was not really missed.
I respectfully beg to differ.
> Morphic also had
> huge implications and bugs.
Actually, in my eyes as a programmer, Traits have much greater
implication than Morphic. Traits might improve everything, not just UI
and glue code.
> If you play with the Traits, like it (or
> not), send a mail. Otherwise read the Traits paper(s). There is lots of
> explanations there, such as why we don't fall in the trap of multiple
> inheritance, what exactly was gained when refactoring and what you can
> do when you use Traits to build new things.
I was going to complain some more about wanting an example of a
refactoring, then I skimmed the papers. So, for
non-language-researchers, to see how the collections turned out, begin
with section 7 in
and go for the details in -
Having read those, it important to say that However wonderful a language
technology may be to those that use it, new language features can be a
real obstacle to those who have not previously met it. A short tutorial
on how to a really short portion of the collections refactoring would
really help alot.
BTW, this link in the prototype page is broken
are we missing out on something, or is it the ECOOP paper?
More information about the Squeak-dev