[ANN] new update 6707
Martin Wirblat
sql.mawi at t-link.de
Mon Dec 19 23:00:13 UTC 2005
Adrian Lienhard wrote:
>
> Since there were no concernes raised and there is quite some pending
> work to get pushed in 3.9a, we decided to make the traitss changes go
> in now.
>
No concerns were raised??? Probably you don't mean what you say ;-)
Here are the concerns (again):
- 98% of all Squeak users want Tweak and not Traits. Imagine the amount
of anger that will come over you if they find out that Traits delays
Tweak or makes it even "somehow" persistently unavailable to them.
- Smalltalk's class hierarchy and inherited inst vars help programmers
to understand their programs. The "dualistic mode of modeling" with
traits and classes may proof being counterproductive to Smalltalk
programming. The examples so far were not encouraging. In short, using
Traits may result too often in code that is harder to understand, to
write and to maintain - the opposite of what is claimed. Remember, not
only skilled researchers use their own invention but many more mere
mortal programmers *try to achieve* something.
You produced a current standalone image with Traits. Fine! Please let
people try it now. Let those who are interested find out about Traits
themselves.
This takes time, and that is OK. Any trial of ramming it violently into
the main Squeak at this time looks only as if you researchers fear
yourselves that in the end it comes out that Traits in its current form
is not the right thing for Smalltalk.
Traits may have potential as it is now, or maybe it needs some changes.
Perhaps a more radical altered language is appropriate. Let us find out
with time!
Regards,
Martin
More information about the Squeak-dev
mailing list
|