[Modules] Name space semantics was: Re: Some of my thoughts

Bijan Parsia bparsia at email.unc.edu
Fri Aug 17 15:52:36 UTC 2001


On Thu, 16 Aug 2001, Stephen Pair wrote:

[snip[
> Imagine working in Squeak just as you do today, only with no fear...

Er...perhaps foolishly...I working Squeak just as I do today with no
fear. :)

> and
> knowing that anything you change can be zapped into another image and
> run flawlessly, no matter what else is (or isn't) loaded into that other
> image.  That's the way it should be.

That sounds great. *Wildly* implausible, but great.

Something has to give somewhere. Either I must change how I work, or there
will be situations where I must tweak something. Actually, how I work
*now* is by tweaking something when necessary.

*No* module system, *no* componant system, *no* type system has, as far as
I can tell, solved these problems seamlessly. Ever. Java most certainly
not. (Yes, you can point to specific scenarios where Java's package system
made things go "smoother" (if you look locally) than what you *might* have
to do in Squeak....but the global experience, and the *common* experience,
is of incompatibility, hackery, pain, really.long.urly.type.wankery, and
classpath *HELL*.)

In the better systems, i.e., the reflective, introspective ones, modules
and components are new constructs. *Programming* constructs. Objects, if
you will. We'll have to learn all about them. We should *want* to learn
all about them.

So, perhaps we should not just hope to alleviate some pain, but to
generate some pleasure :)

Cheers,
Bijan Parsia.





More information about the Squeak-dev mailing list