Whoa!  After years of arguing that we should be using Dictionary's and #=,
instead of IdentityDictionary's and #==, someone pops up even _more_
sympathetic to it than crazy ol'e me, LOL!

On Fri, Apr 24, 2020 at 12:08 PM Marcel Taeumel <marcel.taeumel at hpi.de>
wrote:

> > But you already know that this is a class name...
>
> That's not the point. I am arguing for simplicity and discoverability. You
> arguing in favor of some implementation detail for a performance gain.
>

I'm not.  For me, it's the fact that the class names managed by Smalltalk
really are known from the outside to be semantically symbolic, distinct and
canonical.  Like ENUM values.  Not "names" of undistinct objects like
"Fred" .

But, your personal preference, Marcel, is much more palatable to me than
the places we employ IdentityDictionary's and #== because we "know" we're
using Symbols in underneath.


> There should be a #classAt: if you want the client to specify an
> identifier.
>

Yes, but having both is overkill and #classNamed: is too legacy to rename.
The fact this can be suggested this points to the symbolic nature of the
names!  ;-)

Best,
  Chris
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20200425/eef6e67f/attachment.html>