[vwnc] capitalized strings
Locked
 
|
While composing the other message, I found myself once again searching
for a method capitalizing a string, i.e. ensuring that the first letter is upper case. This is what I found: AbstractWorkspace and Tools.FileTools.ParcelFileViewer both implement a private method "capitalized: aString" with cosmetic differences. A rewriter search for `string at: 1 put: `string first asUppercase returned several places: Refactory.Browser.CompositeLintRule class>>ruleFor:protocol: Refactory.Browser.Prerequisite>>printOn: Tools.FileTools:ParcelFileViewer>>capitalized: CodeWriter>>storeSourceHeader InteractiveCompilerErrorHandler>>declareNewClass:from: ExternalInterfaceBuilder>>updateExternalLabel ExternalInterfaceBuilder>>updateCategorySelected ManualWriter>>reportExtendedDefinitionFor:on: Net.HeaderField>>canonicalFieldName ProfileOutlineBrowser>>printReport:on: The package Squeak-Extensions defines capitalized for Symbol and String, using (`string at: 1) instead of `string first, so I checked for that, too: ICC.ICC1Utils class>>validateClassName:confirm:warn: ICC.IncrementalTypes.IT1FullType>>capitalizedFor: String>>capitalized SystemUtils class>>validateClassName:confirm:warn: These are "just the facts" from my current development image. Pros and Cons re adding String>>capitalized to the base system? Thanks, s. _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
At 02:18 PM 6/13/2008, Stefan Schmiedl wrote:
of discovering a plethora of mutants, scattered about, whereas he (and we) might well prefer something else. ...details snipped... >These are "just the facts" from my current development image. >Pros and Cons re adding String>>capitalized to the base system? The name #capitalized strikes me as ambiguous - does it update in place, or answer a value? String>>capitalize " mutable Strings can self modify, (hence the verb) but these's another problem... " self at: 1 put: (self at: 1) asUppercase String>>asCapitalized " One could over-optimize this, and answer self - iff already capitalized, but that lacks important properties. Could instead answer a copy for that case, which lacks fewer of said properties, but is still lacking. This is better: (still has a problem, though) " ^self copyReplaceFrom: 1 to: 1 with: (self at: 1) asUppercase So TSTTCPW appears to me to be String>>asCapitalized ^self first asUppercase, (self copyFrom: 2) Note: I'm assuming that #, is implemented on Character, and that aString respondsTo: #copyFrom:, neither of which is true, but should be. Regards, -cstb _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
In reply to this post by Stefan Schmiedl
Am 13.06.2008 um 23:18 schrieb Stefan Schmiedl: > While composing the other message, I found myself once again searching > for a method capitalizing a string, i.e. ensuring that the first > letter > is upper case. String >> asMixedcase | copy | self size == 0 ifTrue:[^self]. copy := self species new: self size. copy at: 1 put: (self at: 1) asUppercase. self size > 1 ifTrue:[ 2 to: self size do: [:index | copy at: index put: (self at: index) asLowercase]]. ^copy Andre _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
Re: [vwnc] capitalized strings
|
|
Hi Andre,
thanks for your code, it made me think! On Sat, 14 Jun 2008 14:53:27 +0200 Andre Schnoor <[hidden email]> wrote: > > Am 13.06.2008 um 23:18 schrieb Stefan Schmiedl: > > > While composing the other message, I found myself once again searching > > for a method capitalizing a string, i.e. ensuring that the first > > letter is upper case. > > String >> asMixedcase > | copy | > self size == 0 ifTrue:[^self]. > copy := self species new: self size. > copy at: 1 put: (self at: 1) asUppercase. > self size > 1 ifTrue:[ > 2 to: self size do: > [:index | copy at: index put: (self at: index) asLowercase]]. > ^copy hmm... you're forcing the rest of the string to lower case. I have not yet needed to do this, but if I would have to, I'd lowercase the whole string and only uppercase (are those verbs?) the first character. Do you have a specific reason why you're implementing everything by yourself? Using stuff that is already there, the code would shorten to: CharacterArray>>asMixedcaseStefan1 | copy | self isEmpty ifTrue: [^self]. copy := self asLowercase. copy at: 1 put: copy first asUppercase. ^copy You could even get rid of the local variable: CharacterArray>>asMixedcaseStefan2 self isEmpty ifTrue: [^ self]. ^ (self asLowercase) at: 1 put: self first asUppercase; yourself Ok, now I've got me curious: | t1 | [ t1 := 'frischGemauertInDerErdenStehtDieFormAusLehmGebrannt' asMixedcaseAndre ] benchmarkMicrosecondsToRunRepetitions: 1000000 2.04113 1.65703 2.06479 1.85676 1.84561 1.69418 1.69391 | t2 | [ t2 := 'frischGemauertInDerErdenStehtDieFormAusLehmGebrannt' asMixedcaseStefan1 ] benchmarkMicrosecondsToRunRepetitions: 1000000 1.96859 1.9153 1.95549 1.96726 1.70966 1.91922 2.12955 | t3 | [ t3 := 'frischGemauertInDerErdenStehtDieFormAusLehmGebrannt' asMixedcaseStefan2 ] benchmarkMicrosecondsToRunRepetitions: 1000000 1.7299 1.73017 1.6999 1.72513 1.73166 1.73661 1.73176 1.72689 So it's not the concern for runtime that made you write this. Historical reasons? Just making Smalltalk, s. _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
Re: [vwnc] capitalized strings
|
|
In reply to this post by Stefan Schmiedl
All of the methods described below assume that the string has at least one character, and some assume that it has at least two characters. Do you really want to throw exceptions on empty and single-character strings? Cheers, Alan ----- Original Message ---- From: cstb <[hidden email]> To: Stefan Schmiedl <[hidden email]> Cc: "[hidden email]" <[hidden email]> Sent: Friday, June 13, 2008 10:11:04 PM Subject: Re: [vwnc] capitalized strings At 02:18 PM 6/13/2008, Stefan Schmiedl wrote: of discovering a plethora of mutants, scattered about, whereas he (and we) might well prefer something else. ...details snipped... >These are "just the facts" from my current development image. >Pros and Cons re adding String>>capitalized to the base system? The name #capitalized strikes me as ambiguous - does it update in place, or answer a value? String>>capitalize " mutable Strings can self modify, (hence the verb) but these's another problem... " self at: 1 put: (self at: 1) asUppercase String>>asCapitalized " One could over-optimize this, and answer self - iff already capitalized, but that lacks important properties. Could instead answer a copy for that case, which lacks fewer of said properties, but is still lacking. This is better: (still has a problem, though) " ^self copyReplaceFrom: 1 to: 1 with: (self at: 1) asUppercase So TSTTCPW appears to me to be String>>asCapitalized ^self first asUppercase, (self copyFrom: 2) Note: I'm assuming that #, is implemented on Character, and that aString respondsTo: #copyFrom:, neither of which is true, but should be. Regards, -cstb _______________________________________________ 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 |
Re: [vwnc] capitalized strings
|
|
In reply to this post by Stefan Schmiedl
On Jun 14, 2008, at 12:41 PM, Stefan Schmiedl wrote: > Hi Andre, > > thanks for your code, it made me think! > > On Sat, 14 Jun 2008 14:53:27 +0200 > Andre Schnoor <[hidden email]> wrote: > >> >> Am 13.06.2008 um 23:18 schrieb Stefan Schmiedl: >> >>> While composing the other message, I found myself once again >>> searching >>> for a method capitalizing a string, i.e. ensuring that the first >>> letter is upper case. >> >> String >> asMixedcase >> | copy | >> self size == 0 ifTrue:[^self]. >> copy := self species new: self size. >> copy at: 1 put: (self at: 1) asUppercase. >> self size > 1 ifTrue:[ >> 2 to: self size do: >> [:index | copy at: index put: (self at: index) asLowercase]]. >> ^copy > > hmm... you're forcing the rest of the string to lower case. I have not > yet needed to do this, but if I would have to, I'd lowercase the whole > string and only uppercase (are those verbs?) the first character. > > Do you have a specific reason why you're implementing everything > by yourself? Using stuff that is already there, the code would shorten > to: > > CharacterArray>>asMixedcaseStefan1 > | copy | > self isEmpty ifTrue: [^self]. > copy := self asLowercase. > copy at: 1 put: copy first asUppercase. > ^copy > > You could even get rid of the local variable: > > CharacterArray>>asMixedcaseStefan2 > self isEmpty ifTrue: [^ self]. > ^ (self asLowercase) > at: 1 put: self first asUppercase; > yourself I'd forgo worrying about speed until someone showed me it mattered and do: CharacterArray>>raiseFirst (self first: 1) asUppercase , (self allButFirst: 1) It expressed pretty clear (imho) and doesn't have to worry about empty collections. If I was having a moment of Dolphin envy/nostalgia, I might be tempted to implement CharacterArray>>raiseFirst | block | block := [:toRaise | block := [:toPass | toPass]. toRaise asUppercase]. ^self collect: [:each | block value: each] I'm being silly here on the second, but it's a fun thinker. -- 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 |
|
|
In reply to this post by Alan Darlington
At 12:49 PM 6/14/2008, Alan Darlington wrote:
All of the methods described below assume that the string has at least one character, Correct - the precondition for all of them is [self notEmpty]. and some assume that it has at least two characters. No - you're guessing a different implementation for #copyFrom: than I intended - which is fair, since I didn't specify it. I implement it as an alias for #allButFirst:. Do you really want to throw exceptions on empty and single-character strings? Yes - I do. Since #capitalized is defined in terms of #first, empty strings refuse capitalization, thus matching the exception-behavior of that which they depend on, figuratively speaking. Seemed only fair to make it literally true, as well. Cheers, -cstb Cheers, _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
In reply to this post by Travis Griggs-3
At 04:05 PM 6/14/2008, Travis Griggs wrote:
>On Jun 14, 2008, at 12:41 PM, Stefan Schmiedl wrote: > >> Hi Andre, >> >> thanks for your code, it made me think! +1. >> >> On Sat, 14 Jun 2008 14:53:27 +0200 >> Andre Schnoor <[hidden email]> wrote: >> >>> >>> Am 13.06.2008 um 23:18 schrieb Stefan Schmiedl: >>> >>>> While composing the other message, I found myself once again >>>> searching >>>> for a method capitalizing a string, i.e. ensuring that the first >>>> letter is upper case. >>> >>> String >> asMixedcase >>> | copy | >>> self size == 0 ifTrue:[^self]. >>> copy := self species new: self size. >>> copy at: 1 put: (self at: 1) asUppercase. >>> self size > 1 ifTrue:[ >>> 2 to: self size do: >>> [:index | copy at: index put: (self at: index) asLowercase]]. >>> ^copy >> >> hmm... you're forcing the rest of the string to lower case. I have not >> yet needed to do this, but if I would have to, I'd lowercase the whole >> string and only uppercase (are those verbs?) the first character. >> >> Do you have a specific reason why you're implementing everything >> by yourself? Using stuff that is already there, the code would shorten >> to: >> >> CharacterArray>>asMixedcaseStefan1 >> | copy | >> self isEmpty ifTrue: [^self]. >> copy := self asLowercase. >> copy at: 1 put: copy first asUppercase. >> ^copy >> >> You could even get rid of the local variable: >> >> CharacterArray>>asMixedcaseStefan2 >> self isEmpty ifTrue: [^ self]. >> ^ (self asLowercase) >> at: 1 put: self first asUppercase; >> yourself > >I'd forgo worrying about speed until someone showed me it mattered and >do: > >CharacterArray>>raiseFirst >(self first: 1) asUppercase , (self allButFirst: 1) Hey - that first: bit is a nice touch, even though it produces such horribly pragmatic results, rubidiomatically, some might say. Can't allow it. Otherwise, you're advocating almost the same solution, for about the same reason, excepting the exception. Except that, having seen yours, mine becomes >>capitalized ^self first asUppercase, self allButFirst Yes, I know - I started out claiming #capitalized was an ambiguous name. Unfortunately, that was Fd Up BS, and also wrong. ;-) >>asProperNoun ^self first asUppercase, self allButFirst asLowercase >>capitalizedOr: a0block ^self isEmpty ifTrue: [a0block value] ifFalse: [self capitalized] >>asMixedCase ^(self allSatisfy: [:ch| ch isUppercase]) ifTrue: [self first asLowercase, self allButFirst] ifFalse: [self capitalized] Cheers, -cstb _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
In reply to this post by Travis Griggs-3
> On Jun 14, 2008, at 12:41 PM, Stefan Schmiedl wrote: > >> Do you have a specific reason why you're implementing everything >> by yourself? Using stuff that is already there, the code would >> shorten >> to: .... I didn't mean the code to be especially elegant or short. Just copied it from my current image for instant use. It's simple, handles the case of empty strings and doesn't create much garbage. IIRC, I once copied and modified it from somewhere else in the String hierarchy. Did not much think about it. Short and elegant expressions look nice (and Smalltalk is tempting in this regard), but they tend to incur ineffeciency (garbage, unnecessary level of abstraction). Below a certain threshold of bytecode size, it can sometimes be more effective to do things directly. Andre _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
Re: [vwnc] capitalized strings
|
|
On Sun, 15 Jun 2008 14:06:40 +0200
Andre Schnoor <[hidden email]> wrote: > > > On Jun 14, 2008, at 12:41 PM, Stefan Schmiedl wrote: > > > >> Do you have a specific reason why you're implementing everything > >> by yourself? Using stuff that is already there, the code would > >> shorten > >> to: .... > > I didn't mean the code to be especially elegant or short. Just copied > it from my current image for instant use. It's simple, handles the > case of empty strings and doesn't create much garbage. IIRC, I once > copied and modified it from somewhere else in the String hierarchy. > Did not much think about it. Thanks for the explanation. I'm often wondering about the why of things for fear of missing something. Copy and tweak is a quite success technique for problem solving. > Short and elegant expressions look nice (and Smalltalk is tempting in > this regard), but they tend to incur ineffeciency (garbage, > unnecessary level of abstraction). Below a certain threshold of > bytecode size, it can sometimes be more effective to do things directly. That's what I'm often wondering, so imagine my surprise when there was no runtime difference between your "direct manipulation" and reusing the available methods. Just making Smalltalk, s. _______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
|
|
In reply to this post by Travis Griggs-3
Hi,
I have on my VW image the method that was in VSE. I like the name (but I'm using it from at least 15 years...) Travis, the name raiseFirst is one I'll never browse for if I need to find if something that make a change in the sense of capitalizing or converting to uppercase is already there. Probably this is because English is not my first language, but seems to me not the right name. In any case, I'll probably search the system (browse implementors...) for *Uppercase* or *Capitalize*, surely never for *Raise* For the History .... Here is the version and comment from original VSE implementation: (the one living on my VW image, lazily copied..) asFirstUppercase "(public) Answer a String containing the receiver with first character in upper-case." | answer | (self size < 2) ifTrue: [^self asUppercase]. answer := self copyFrom: 1 to: (self size). answer at: 1 put: ( self at: 1 ) asUppercase. ^answer ciao Giorgio On Sun, Jun 15, 2008 at 1:05 AM, Travis Griggs <[hidden email]> wrote:
_______________________________________________ vwnc mailing list [hidden email] http://lists.cs.uiuc.edu/mailman/listinfo/vwnc |
«
Return to VisualWorks
|
1 view|%1 views
| Free forum by Nabble | Edit this page |