[vwnc] [7.6][BUG] Dictionary>>groupedBy:

(Dictionary new
        at: 1 put: 'abc';
        at: 2 put: 'abcd';
        at: 3 put: 'abcde';
        yourself)
                groupedBy: [:ea | ea size]

... blows up as per attached, whereas I would have expected it to return
a dictionary with values re-keyed using the passed block?

Cheers!

-Boris

--
+1.604.689.0322
DeepCove Labs Ltd.
4th floor 595 Howe Street
Vancouver, Canada V6C 2T5
http://tinyurl.com/r7uw4

[hidden email]

CONFIDENTIALITY NOTICE

This email is intended only for the persons named in the message
header. Unless otherwise indicated, it contains information that is
private and confidential. If you have received it in error, please
notify the sender and delete the entire message including any
attachments.

Thank you.


_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

On Jul 28, 2008, at 8:52 AM, Boris Popov wrote:


Would you expect this:

(Dictionary
        with: (3 -> (Dictionary with: 1 -> 'abc'))
        with: (4 -> (Dictionary with: 2 -> 'abcd'))
        with: (5 -> (Dictionary with: 3 -> 'abcde'))

or

(Dictionary
        with: (3 -> (OrderedCollection with: 'abc'))
        with: (4 -> (OrderedCollection with: 'abcd'))
        with: (5 -> (OrderedCollection with: 'abcde'))


By default any grouping operation creates the second first. But then  
there's a 2nd part of the method that tries to be tricky. It basically  
tries to make the "groups" be the same type as the receiver. The idea  
being that if you do:

(Bag withAll: #('abc' 'abcd' 'abcde')) groupedBy: #size, you get

(Dictionary
        with: (3 -> (Bag with: 'abc'))
        with: (4 -> (Bag with: 'abcd'))
        with: (5 -> (Bag with: 'abcde'))

I'm not sure what I think of the second stage trickery here. Part of  
me just thinks the groupedBy: should always do the base thing  
(OrderedCollections). The other half says that if we're going to make  
collect: return a dictionary, then we should do as #1 above does. We  
definitely shouldn't blow up, that's Bad (tm).

A less efficient way of writing the method, but one that uses more of  
the existing services, rather than special casing the return value of  
species class, would be something like:

groupedBy: aBlock
        | keys |
        keys := Set withAll: (self collect: aBlock).
        ^Dictionary
                withAll:
                        (keys
                                collect:
                                        [:eachKey | eachKey -> (self select: [:each | eachKey = (aBlock  
value: each)])])

This would produce the first result, and work for each receiver types  
in terms of it's collect:/select: semantics without having to futz  
around with species types, or recreating collections on a sometimes  
basis. It is however, about 2.5 x slower. Except for the current  
Dictionary case, which case it's infinitely faster because the  
original fails.

--
Travis Griggs
Objologist
"Some people are like slinkies, not really good for much, but they can  
bring a smile to your face when you push them down the stairs."

_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

Travis,

I would expect a second result, actually. If I have a collection of
widgets keyed by integers (as is the case for arrays) or by gadgets (as
is the case for dictionaries), grouping either one should be producing a
dictionary where key is the result of evaluating a block and value is a
collection of all items matching that result. I don't care about the
original keys in either case...

-Boris

--
+1.604.689.0322
DeepCove Labs Ltd.
4th floor 595 Howe Street
Vancouver, Canada V6C 2T5
http://tinyurl.com/r7uw4

[hidden email]

CONFIDENTIALITY NOTICE

This email is intended only for the persons named in the message
header. Unless otherwise indicated, it contains information that is
private and confidential. If you have received it in error, please
notify the sender and delete the entire message including any
attachments.

Thank you.

> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]] On
Behalf select: [:each
_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

On Jul 28, 2008, at 12:07 PM, Boris Popov wrote:

So what about the Bag case, what would you expect there?

--
Travis Griggs
Objologist
"I think that we should be men first, and subjects afterward." - Henry  
David Thoreau



_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

Yea, I see what you're getting at; you want values to always be the same
type of a collection that you start with... In that case I guess I could
always do,

aDictionary values groupedBy: [:ea | ea size]

in cases where I want to re-key the whole thing, but then you'd still
want to fix the error for direct groupedBy: sends.

-Boris

--
+1.604.689.0322
DeepCove Labs Ltd.
4th floor 595 Howe Street
Vancouver, Canada V6C 2T5
http://tinyurl.com/r7uw4

[hidden email]

CONFIDENTIALITY NOTICE

This email is intended only for the persons named in the message
header. Unless otherwise indicated, it contains information that is
private and confidential. If you have received it in error, please
notify the sender and delete the entire message including any
attachments.

Thank you.

> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]] On
Behalf
_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

On Jul 28, 2008, at 12:07 PM, Boris Popov wrote:

Restating, just to make sure, you're looking for a Dictionary  
transform function, which operates on the keys, rather than on values?  
Is that correct?

i.e.

(Dictionary with: 4 -> 5 with: 3 -> 7) borisFunctor: #asFloat -->  
(Dictionary with: 4.0 -> 5 with: 3.0 -> 7)

Or am I still being dense?

You discovered a bug in groupedBy: either way :)

--
Travis Griggs
Objologist
"I think that we should be men first, and subjects afterward." - Henry  
David Thoreau



_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

Umm, I wasn't thinking that far, all I was trying to do is re-key my
dictionary using another parameter. Say you have a dictionary of widgets
keyed by price and you want to get a dictionary of widgets keyed by
color. My natural instinct would be to say,

$10.00 -> #(aWidget bWidget)
$20.00 -> #(cWidget)

myWidgetsByPrice groupedBy: #color

#blue -> #(aWidget)
#green -> #(bWidget)
#yellow -> #(cWidget)

whereas it appears I really need to do,

myWidgetsByPrice values groupedBy: #color

-Boris

--
+1.604.689.0322
DeepCove Labs Ltd.
4th floor 595 Howe Street
Vancouver, Canada V6C 2T5
http://tinyurl.com/r7uw4

[hidden email]

CONFIDENTIALITY NOTICE

This email is intended only for the persons named in the message
header. Unless otherwise indicated, it contains information that is
private and confidential. If you have received it in error, please
notify the sender and delete the entire message including any
attachments.

Thank you.

> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]] On
Behalf
_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

Actually, I would also need to fold the values,

(((Dictionary new)
        at: 10 put: #('a' 'bb');
        at: 20 put: #('ccc');
        yourself) values fold: [:a :b | a , b]) groupedBy: [:ea | ea
size]

I might just end up with a specialized implementation for re-keying and
let you fix the default implementation of groupedBy: on Dictionaries.

-Boris

--
+1.604.689.0322
DeepCove Labs Ltd.
4th floor 595 Howe Street
Vancouver, Canada V6C 2T5
http://tinyurl.com/r7uw4

[hidden email]

CONFIDENTIALITY NOTICE

This email is intended only for the persons named in the message
header. Unless otherwise indicated, it contains information that is
private and confidential. If you have received it in error, please
notify the sender and delete the entire message including any
attachments.

Thank you.

> -----Original Message-----
> From: Boris Popov
> Sent: Monday, July 28, 2008 1:03 PM
> To: 'Travis Griggs'; VWNC List
> Subject: RE: [vwnc] [7.6][BUG] Dictionary>>groupedBy:
>
> Umm, I wasn't thinking that far, all I was trying to do is re-key my
> dictionary using another parameter. Say you have a dictionary of
widgets
> keyed by price and you want to get a dictionary of widgets keyed by
color. of
> > > widgets keyed by integers (as is the case for arrays) or by
gadgets
> > > (as
> > > is the case for dictionaries), grouping either one should be
> > > producing a
> > > dictionary where key is the result of evaluating a block and value
> > > is a
> > > collection of all items matching that result. I don't care about
the
> > > original keys in either case...
> >
> > Restating, just to make sure, you're looking for a Dictionary
> > transform function, which operates on the keys, rather than on
values? Henry
> > David Thoreau
> >
> >
> >
> > _______________________________________________
> > vwnc mailing list
> > [hidden email]
> > http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

Boris Popov wrote: I don't think that'll be enough. It'll make a collection of the results
that match the new key - so if your original dictionary had values that
are collections, the new dictionary will have values that are
collections of collections.

It sort of seems like you need a specific API for your problem here...

Michael
_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

Boris, I did submit an AR (54875) for this. Thanks


On Jul 28, 2008, at 8:52 AM, Boris Popov wrote:

--
Travis Griggs
Objologist
"I think that we should be men first, and subjects afterward." - Henry  
David Thoreau



_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

Boris,

Would this do for your case?

---

Collection>>allElementsGroupedBy: a1block
        | regrouped |
        regrouped := Dictionary new.
        self allElementsDo:
                [:each|
                 key := a1block value: each.
                 regrouped at: key add: each.
                ].
        ^regrouped

allElementsDo: a1block
        ^self do: [:each| each allElementsDo: a1block]

Object>>allElementsDo: a1block
        ^a1block value: selfTravis,

---

Travis,

For #groupedBy:, your examples
look like they answer something I 'd
call a thesaurous.  Buta bea carfull,
ok, becausea thesa U rous', are, asa
you know,  asometimes quitea dangerous...

Regards,

-cstb

---

Collection>>groupedBy: a1block
        | new |
        grouped := Thesaurus of: self class.
        self elementsDo:
                [:each| | key |
                 key := a1block value: each.
                 grouped at: key add: each
                ].
        ^grouped

>>addElement: anObject
        ^self add: anObject

>>elementsDo: a1block
        ^self do: a1block

Dictionary>>addElement: anAssociation
        self addAssociation: anAssociation

>>elementsDo: a1block
        ^self associationsDo: a1block

---

Here's a minimal thesaurus, for #groupedBy:

---

Collection
        subclass: Thesaurus
        iVars: 'newEntry entryClass'

Thesaurus class>>of: aCollectionClass
        ^self new
                setEntryClass: aCollectionClass
                ; yourself

>>initialize
        super initialize.
        self setEntryClass: OrderedCollection

>>setEntryClass: aCollectionClass
        newEntry := [entryClass new].
        ^entryClass := aCollectionClass

>>at: key add: anObject
        | entry |
        entry := self at: key ifAbsentPut: newEntry.
        ^entry addElement: anObject


Regards,

-cstb


At 12:59 PM 7/31/2008, Travis Griggs wrote:

_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc