I stumbled upon

The following two functions:
capitalized
        "Return a copy with the first letter capitalized"
        | cap |
        self isEmpty ifTrue: [ ^self copy ].
        cap := self copy.
        cap at: 1 put: (cap at: 1) asUppercase.
        ^ cap

and
withFirstCharacterDownshifted
        "Return a copy with the first letter downShifted (in lower case)"
       
        | answer |
        self ifEmpty: [ ^ self copy ].
        answer := self copy.
        answer at: 1 put: answer first asLowercase.
        ^ answer

Well AFAIKT it should be first in the function (according to SLIME)

But as one can easily see it's redundant code. How about a replacement
like:
withFirstLetterTransformed: aTransformBlock
        | result |
        self isEmpty ifTrue: [^ self copy ].
        result := self copy.
        result at: 1 put: (aTransformBlock value: result first ).
        ^ result
       
the above fucntions would get:
capitalized
        "Return a copy with the first letter capitalized"
        ^ self  withFirstLetterTransformed: [:value | value asUppercase ].

and
withFirstCharacterDownshifted
        "Return a copy with the first letter downShifted (in lower case)"
       
        ^ self withFirstLetterTransformed: [:value | value asLowercase ].

Well I also know that capitalized is a very "old" thing but the second
name is more specific. So what about:
withFistCharacterUpshifted .... as counterpart?


There's another thing which probably is incongruent:
we have first for accessing the first element of any collection
but first: is not for setting the first element.

first:n yields the first n Elements of a collection.

That's not a good name then.
firstElements:n would be a better name IMHO.

And then we could have
first: someValue
and so we would have in the end:

result first: (aTranformBlock value: result first).


would change the first Element of "any" collection. AFAIKT Lisp has such
a convention...

my first changes to not to any harm to existing Code, but of course
changing the meaning of first: would....

Regards
Friedrich

Hi Friedrich,

I had a similar problem in Dolphin Smalltalk some days ago.
Dolphin has String>>capitalized:
capitalized
        "Answer a <readableString> which is a copy of the receiver but with
        the first character converted to its uppercase equivalent."

        | answer |
        (answer := self basicCopy) notEmpty ifTrue: [answer at: 1 put: self first asUppercase].
        ^answer

I was in a need of the opposite (I called it decapitalized) and ended with
String>>withFirstCharacterTransformedWith: aMethodSymbol
        "Answer a <readableString> which is a copy of the receiver but with
        the first character converted by performing aMethodSymbol on it."

        | answer |
        (answer := self basicCopy) notEmpty ifTrue: [answer at: 1 put: (self first perform: aMethodSymbol)].
        ^answer

Comparing my solution with your first approach I like mine better :)
But following your arguments about consistency I like your final approach best.
Alas IMO it should be initiated by the ANSI Smalltalk team (is it still alive?) as it would be
nice to have this in all dialects.

Best regards
Andreas


Am 29.03.2013 um 07:55 schrieb Friedrich Dominicus <[hidden email]>:

On Mar 29, 2013, at 10:28 AM, Andreas Wacknitz <[hidden email]> wrote:
>
>
> Comparing my solution with your first approach I like mine better :)
> But following your arguments about consistency I like your final approach best.
> Alas IMO it should be initiated by the ANSI Smalltalk team (is it still alive?)

ANSI is dead since a long time.

        Marcus

>
>
> the above fucntions would get:
> capitalized
> "Return a copy with the first letter capitalized"
> ^ self  withFirstLetterTransformed: [:value | value asUppercase ].
>
> and
> withFirstCharacterDownshifted

withFirstCharacterLowercase?


> "Return a copy with the first letter downShifted (in lower case)"
>
> ^ self withFirstLetterTransformed: [:value | value asLowercase ].
>
> Well I also know that capitalized is a very "old" thing but the second
> name is more specific. So what about:
> withFistCharacterUpshifted .... as counterpart?

does not fly :)


> we have first for accessing the first element of any collection
> but first: is not for setting the first element.
>
> first:n yields the first n Elements of a collection.
>
> That's not a good name then.
> firstElements:n would be a better name IMHO.

indeed.


Did you check what are the senders of first:?

STef

FDominicus wrote

Well I also know that capitalized is a very "old" thing but the second
name is more specific. So what about:
withFistCharacterUpshifted .... as counterpart?

I see what you mean... Apple's dictionary says that capitalized could mean "all caps" or "first letter capitalized", so I guess it's ambiguous.

For #withFirstCharacterDownshifted, isn't downshifted misleading? It only works on letters. For instance, it doesn't change $: to $;. I would change #withFirstCharacterDownshifted->#withFirstCharacterLowercase.

And if I changed #capitalized, it would be to something like #withFirstCharacterCapitalized (or uppercase). Also, the method you extracted uses "letter" instead of "character". They should all be uniform.

Sean P. DeNigris wrote:
That could just be #withFirstCapitalized - you already know you are
dealing with characters (?) ...

stephane ducasse <[hidden email]> writes:
Well no, the name in the Pharo image is withFirstCharacterDownshifted
withFirstCharacterInLowercase?
withFirstCharacterInUppercase?
Well that might be. I just wrote what I found and I was thinking about
unifiying it a bit...

>
> Did you check what are the senders of first:?
No, but I checked today. There are 35 senders of it in my Image (pharo
2-0 one click)

Regards
Friedrich

"Sean P. DeNigris" <[hidden email]> writes:
Oh I agree, I just mentioned what I found in my image. And I's just
proposed to change that.

The name are not from me but what there is in the current image.
So there is capitalize and there is withFirstCharacterDownshifted. And
the have a very similiar implementation which I just put into another
message.

But I've not idea on how to make a proposal, but in this way. Sending it
to this mailing list...

Regards
Friedrich