Smalltalk › Pharo › Pharo Smalltalk Developers

about threeWayCompareTo:

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

5 messages Options

ducasse

about threeWayCompareTo:

I see

threeWayCompareTo: aString
"Do a three-way comparison between the receiver and anotherObject, returning
-1 if self < anotherObject
0 if self = anotherObject
1 if self > anotherObject
This assumes a total order in accordance with the mathematical law of trichotomy.
See also: http://en.wikipedia.org/wiki/Three-way_comparison"

^ (self compare: self with: aString collated: AsciiOrder) - 2

And I thought that we got a new primitive returning -101 directly
Is it not the case?

Stef

String>>compare: string1 with: string2 collated: order

"'abc' = 'abc' asWideString >>> true"
"'abc' asWideString = 'abc' >>> true"
"(ByteArray with: 97 with: 0 with: 0 with: 0) asString ~= 'a000' asWideString >>> true"
"('abc' sameAs: 'aBc' asWideString) >>> true"
"('aBc' asWideString sameAs: 'abc') >>> true"
"('a000' asWideString ~= (ByteArray with: 97 with: 0 with: 0 with: 0) asString) >>> true"
"((ByteArray with: 97 with: 0 with: 0 with: 0) asString sameAs: 'Abcd' asWideString) >>> false"
"('a000' asWideString sameAs: (ByteArray with: 97 with: 0 with: 0 with: 0) asString) >>> false"

(string1 isByteString and: [string2 isByteString]) ifTrue: [
^ ByteString compare: string1 with: string2 collated: order].
"Primitive does not fail properly right now"
^ String compare: string1 with: string2 collated: order

ByteString>>compare: string1 with: string2 collated: order
"Return 1, 2 or 3, if string1 is <, =, or > string2, with the collating order of characters given by the order array."

| len1 len2 c1 c2 |
<primitive: 'primitiveCompareString' module: 'MiscPrimitivePlugin'>

<var: #string1 declareC: 'unsigned char *string1'>
<var: #string2 declareC: 'unsigned char *string2'>
<var: #order declareC: 'unsigned char *order'>

len1 := string1 size.
len2 := string2 size.
1 to: (len1 min: len2) do:
[:i |
c1 := order at: (string1 basicAt: i) + 1.
c2 := order at: (string2 basicAt: i) + 1.
c1 = c2 ifFalse:
[c1 < c2 ifTrue: [^ 1] ifFalse: [^ 3]]].
len1 = len2 ifTrue: [^ 2].
len1 < len2 ifTrue: [^ 1] ifFalse: [^ 3].

Nicolas Cellier

Re: about threeWayCompareTo:

This was WIP, don't know if it stalled...

Le lun. 16 sept. 2019 à 08:52, ducasse <[hidden email]> a écrit :

I see

threeWayCompareTo: aString
"Do a three-way comparison between the receiver and anotherObject, returning
-1 if self < anotherObject
0 if self = anotherObject
1 if self > anotherObject
This assumes a total order in accordance with the mathematical law of trichotomy.
See also: http://en.wikipedia.org/wiki/Three-way_comparison"

^ (self compare: self with: aString collated: AsciiOrder) - 2

And I thought that we got a new primitive returning -101 directly
Is it not the case?

Stef

String>>compare: string1 with: string2 collated: order

"'abc' = 'abc' asWideString >>> true"
"'abc' asWideString = 'abc' >>> true"
"(ByteArray with: 97 with: 0 with: 0 with: 0) asString ~= 'a000' asWideString >>> true"
"('abc' sameAs: 'aBc' asWideString) >>> true"
"('aBc' asWideString sameAs: 'abc') >>> true"
"('a000' asWideString ~= (ByteArray with: 97 with: 0 with: 0 with: 0) asString) >>> true"
"((ByteArray with: 97 with: 0 with: 0 with: 0) asString sameAs: 'Abcd' asWideString) >>> false"
"('a000' asWideString sameAs: (ByteArray with: 97 with: 0 with: 0 with: 0) asString) >>> false"

(string1 isByteString and: [string2 isByteString]) ifTrue: [
^ ByteString compare: string1 with: string2 collated: order].
"Primitive does not fail properly right now"
^ String compare: string1 with: string2 collated: order

ByteString>>compare: string1 with: string2 collated: order
"Return 1, 2 or 3, if string1 is <, =, or > string2, with the collating order of characters given by the order array."

| len1 len2 c1 c2 |
<primitive: 'primitiveCompareString' module: 'MiscPrimitivePlugin'>

<var: #string1 declareC: 'unsigned char *string1'>
<var: #string2 declareC: 'unsigned char *string2'>
<var: #order declareC: 'unsigned char *order'>

len1 := string1 size.
len2 := string2 size.
1 to: (len1 min: len2) do:
[:i |
c1 := order at: (string1 basicAt: i) + 1.
c2 := order at: (string2 basicAt: i) + 1.
c1 = c2 ifFalse:
[c1 < c2 ifTrue: [^ 1] ifFalse: [^ 3]]].
len1 = len2 ifTrue: [^ 2].
len1 < len2 ifTrue: [^ 1] ifFalse: [^ 3].

Esteban A. Maringolo

Re: about threeWayCompareTo:

In reply to this post by ducasse

I haven't followed nor maintained the SortFunctions library, but it
relied heavily on the response of that method, by means of the
"spaceship operator" ( #<=>), I think that Sven later changed such
operator to "threeWayCompareTo:".

I don't know whether the primitive is available now, it wasn't back then.

Regards,

Esteban A. Maringolo

On Mon, Sep 16, 2019 at 3:52 AM ducasse <[hidden email]> wrote:

>
> I see
>
> threeWayCompareTo: aString
> "Do a three-way comparison between the receiver and anotherObject, returning
> -1 if self < anotherObject
> 0 if self = anotherObject
> 1 if self > anotherObject
> This assumes a total order in accordance with the mathematical law of trichotomy.
> See also: http://en.wikipedia.org/wiki/Three-way_comparison"
>
> ^ (self compare: self with: aString collated: AsciiOrder) - 2
>
> And I thought that we got a new primitive returning -101 directly
> Is it not the case?
>
> Stef
>
>
> String>>compare: string1 with: string2 collated: order
>
> "'abc' = 'abc' asWideString >>> true"
> "'abc' asWideString = 'abc' >>> true"
> "(ByteArray with: 97 with: 0 with: 0 with: 0) asString ~= 'a000' asWideString >>> true"
> "('abc' sameAs: 'aBc' asWideString) >>> true"
> "('aBc' asWideString sameAs: 'abc') >>> true"
> "('a000' asWideString ~= (ByteArray with: 97 with: 0 with: 0 with: 0) asString) >>> true"
> "((ByteArray with: 97 with: 0 with: 0 with: 0) asString sameAs: 'Abcd' asWideString) >>> false"
> "('a000' asWideString sameAs: (ByteArray with: 97 with: 0 with: 0 with: 0) asString) >>> false"
>
> (string1 isByteString and: [string2 isByteString]) ifTrue: [
> ^ ByteString compare: string1 with: string2 collated: order].
> "Primitive does not fail properly right now"
> ^ String compare: string1 with: string2 collated: order
>
>
> ByteString>>compare: string1 with: string2 collated: order
> "Return 1, 2 or 3, if string1 is <, =, or > string2, with the collating order of characters given by the order array."
>
> | len1 len2 c1 c2 |
> <primitive: 'primitiveCompareString' module: 'MiscPrimitivePlugin'>
>
> <var: #string1 declareC: 'unsigned char *string1'>
> <var: #string2 declareC: 'unsigned char *string2'>
> <var: #order declareC: 'unsigned char *order'>
>
> len1 := string1 size.
> len2 := string2 size.
> 1 to: (len1 min: len2) do:
> [:i |
> c1 := order at: (string1 basicAt: i) + 1.
> c2 := order at: (string2 basicAt: i) + 1.
> c1 = c2 ifFalse:
> [c1 < c2 ifTrue: [^ 1] ifFalse: [^ 3]]].
> len1 = len2 ifTrue: [^ 2].
> len1 < len2 ifTrue: [^ 1] ifFalse: [^ 3].
>
>
>
>

Nicolas Cellier

Re: about threeWayCompareTo:

This comparison was part of MiscPrimitivePlugin.

MiscPrimitivePlugin is a false good idea:

Smalltalk fallback code = VMMaker (slang) source

Unfortunately, this is an illusion because unlike the C source inclusions in ST/X we don't have dynamic slang compilation!

The consequence is that source code is not managed in VMMaker specific package, but in a dialect specific/version specific package!

Squeak/Pharo/Cuis/Newspeak may all have a different version of String at image side.

This is making VMMaker recipe for code generation more fragile than necessary (and not easily portable to Pharo/Cuis/whatever...).

It seems to me that Clement did start a rewrite of essential primitives into proper plugin...

The plan was then to retract support of MiscPrimitivePlugin after a grace period (after all, these are optional primitives).

Le lun. 16 sept. 2019 à 13:39, Esteban Maringolo <[hidden email]> a écrit :

I haven't followed nor maintained the SortFunctions library, but it
relied heavily on the response of that method, by means of the
"spaceship operator" ( #<=>), I think that Sven later changed such
operator to "threeWayCompareTo:".

I don't know whether the primitive is available now, it wasn't back then.

Regards,

Esteban A. Maringolo

On Mon, Sep 16, 2019 at 3:52 AM ducasse <[hidden email]> wrote:
>
> I see
>
> threeWayCompareTo: aString
> "Do a three-way comparison between the receiver and anotherObject, returning
> -1 if self < anotherObject
> 0 if self = anotherObject
> 1 if self > anotherObject
> This assumes a total order in accordance with the mathematical law of trichotomy.
> See also: http://en.wikipedia.org/wiki/Three-way_comparison"
>
> ^ (self compare: self with: aString collated: AsciiOrder) - 2
>
> And I thought that we got a new primitive returning -101 directly
> Is it not the case?
>
> Stef
>
>
> String>>compare: string1 with: string2 collated: order
>
> "'abc' = 'abc' asWideString >>> true"
> "'abc' asWideString = 'abc' >>> true"
> "(ByteArray with: 97 with: 0 with: 0 with: 0) asString ~= 'a000' asWideString >>> true"
> "('abc' sameAs: 'aBc' asWideString) >>> true"
> "('aBc' asWideString sameAs: 'abc') >>> true"
> "('a000' asWideString ~= (ByteArray with: 97 with: 0 with: 0 with: 0) asString) >>> true"
> "((ByteArray with: 97 with: 0 with: 0 with: 0) asString sameAs: 'Abcd' asWideString) >>> false"
> "('a000' asWideString sameAs: (ByteArray with: 97 with: 0 with: 0 with: 0) asString) >>> false"
>
> (string1 isByteString and: [string2 isByteString]) ifTrue: [
> ^ ByteString compare: string1 with: string2 collated: order].
> "Primitive does not fail properly right now"
> ^ String compare: string1 with: string2 collated: order
>
>
> ByteString>>compare: string1 with: string2 collated: order
> "Return 1, 2 or 3, if string1 is <, =, or > string2, with the collating order of characters given by the order array."
>
> | len1 len2 c1 c2 |
> <primitive: 'primitiveCompareString' module: 'MiscPrimitivePlugin'>
>
> <var: #string1 declareC: 'unsigned char *string1'>
> <var: #string2 declareC: 'unsigned char *string2'>
> <var: #order declareC: 'unsigned char *order'>
>
> len1 := string1 size.
> len2 := string2 size.
> 1 to: (len1 min: len2) do:
> [:i |
> c1 := order at: (string1 basicAt: i) + 1.
> c2 := order at: (string2 basicAt: i) + 1.
> c1 = c2 ifFalse:
> [c1 < c2 ifTrue: [^ 1] ifFalse: [^ 3]]].
> len1 = len2 ifTrue: [^ 2].
> len1 < len2 ifTrue: [^ 1] ifFalse: [^ 3].
>
>
>
>

ducasse

Re: about threeWayCompareTo:

Thanks Nicolas.

We should check because cyril told me that the primitive was implemented but not published in the primitive table.

We will have a look.

On 16 Sep 2019, at 19:42, Nicolas Cellier <[hidden email]> wrote:

This comparison was part of MiscPrimitivePlugin.
MiscPrimitivePlugin is a false good idea:
Smalltalk fallback code = VMMaker (slang) source

Unfortunately, this is an illusion because unlike the C source inclusions in ST/X we don't have dynamic slang compilation!
The consequence is that source code is not managed in VMMaker specific package, but in a dialect specific/version specific package!
Squeak/Pharo/Cuis/Newspeak may all have a different version of String at image side.
This is making VMMaker recipe for code generation more fragile than necessary (and not easily portable to Pharo/Cuis/whatever...).

It seems to me that Clement did start a rewrite of essential primitives into proper plugin...
The plan was then to retract support of MiscPrimitivePlugin after a grace period (after all, these are optional primitives).

Le lun. 16 sept. 2019 à 13:39, Esteban Maringolo <[hidden email]> a écrit :
I haven't followed nor maintained the SortFunctions library, but it
relied heavily on the response of that method, by means of the
"spaceship operator" ( #<=>), I think that Sven later changed such
operator to "threeWayCompareTo:".

I don't know whether the primitive is available now, it wasn't back then.

Regards,

Esteban A. Maringolo

On Mon, Sep 16, 2019 at 3:52 AM ducasse <[hidden email]> wrote:
>
> I see
>
> threeWayCompareTo: aString
> "Do a three-way comparison between the receiver and anotherObject, returning
> -1 if self < anotherObject
> 0 if self = anotherObject
> 1 if self > anotherObject
> This assumes a total order in accordance with the mathematical law of trichotomy.
> See also: http://en.wikipedia.org/wiki/Three-way_comparison"
>
> ^ (self compare: self with: aString collated: AsciiOrder) - 2
>
> And I thought that we got a new primitive returning -101 directly
> Is it not the case?
>
> Stef
>
>
> String>>compare: string1 with: string2 collated: order
>
> "'abc' = 'abc' asWideString >>> true"
> "'abc' asWideString = 'abc' >>> true"
> "(ByteArray with: 97 with: 0 with: 0 with: 0) asString ~= 'a000' asWideString >>> true"
> "('abc' sameAs: 'aBc' asWideString) >>> true"
> "('aBc' asWideString sameAs: 'abc') >>> true"
> "('a000' asWideString ~= (ByteArray with: 97 with: 0 with: 0 with: 0) asString) >>> true"
> "((ByteArray with: 97 with: 0 with: 0 with: 0) asString sameAs: 'Abcd' asWideString) >>> false"
> "('a000' asWideString sameAs: (ByteArray with: 97 with: 0 with: 0 with: 0) asString) >>> false"
>
> (string1 isByteString and: [string2 isByteString]) ifTrue: [
> ^ ByteString compare: string1 with: string2 collated: order].
> "Primitive does not fail properly right now"
> ^ String compare: string1 with: string2 collated: order
>
>
> ByteString>>compare: string1 with: string2 collated: order
> "Return 1, 2 or 3, if string1 is <, =, or > string2, with the collating order of characters given by the order array."
>
> | len1 len2 c1 c2 |
> <primitive: 'primitiveCompareString' module: 'MiscPrimitivePlugin'>
>
> <var: #string1 declareC: 'unsigned char *string1'>
> <var: #string2 declareC: 'unsigned char *string2'>
> <var: #order declareC: 'unsigned char *order'>
>
> len1 := string1 size.
> len2 := string2 size.
> 1 to: (len1 min: len2) do:
> [:i |
> c1 := order at: (string1 basicAt: i) + 1.
> c2 := order at: (string2 basicAt: i) + 1.
> c1 = c2 ifFalse:
> [c1 < c2 ifTrue: [^ 1] ifFalse: [^ 3]]].
> len1 = len2 ifTrue: [^ 2].
> len1 < len2 ifTrue: [^ 1] ifFalse: [^ 3].
>
>
>
>