[Cuis] Dictionary asSet

[Dan Norton]

Comments in line...
On 1 Mar 2015 at 15:06, Nicolas Cellier wrote:
> 2015-03-01 10:52 GMT+01:00 Casey Ransberger
> <casey.obrien.r at gmail.com>:
>     I think it should probably answer a set of associations,
> specifically, and also have a gut 
>     feeling that doing so should be relatively cheap. But don't
> worry so much about what I think. 
>     Look at what the other dialects do, consult book blue, etc...
> 
> 
> Please, no.
> A Dictionary behaves as a Collection of values,  it's just that it
> has special keys for indexing its 
> contents.
> Why should asSet do something different than asArray for example?
> If we want a set of associations, we can just write aDictionary
> associations asSet.
> 
Someone looking at the Terse Guide sees #asOrderedCollection, #asBag, etc. answer the 
values of the dictionary in instances of those classes. So they might expect #asSet to follow 
suit.

In Cuis, Dictionary inherits from Set, unlike Squeak. Seems like Dictionary needs to have its 
own #asSet method.
> 
>     Annoying Casey behavior: can you think of an SUnit test that
> confirms the bug? I'm not just 
>     asking you for a test case, not necessarily, but it would be
> good to define expected behavior 
>     in a way that allows us later to know that it has changed
> inadvertently if it does.
> 
>     The test is more important than the fix, if the code is to be
> the curriculum. 
> 
How about...
asSetTest
	| dx old new |
	dx _ Dictionary new
		at: #foo put: 44; at: #bah put: 55; yourself.
	old _ dx className asSymbol.
	new _ dx asSet className asSymbol.
	self shouldnt: [old = new]
>     
>     On Feb 28, 2015, at 6:45 PM, "Dan Norton"
> <dnorton at mindspring.com> wrote:
> 
>     aDictionary asSet should answer a set. Instead, it answers
> itself (a dictionary).
> 
>     I tried to create a new issue but the "Submit new issue" button
> was disabled.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://jvuletich.org/pipermail/cuis_jvuletich.org/attachments/20150301/e08bc61e/attachment-0004.html>