add std.uni.toCapitalized()

[WalterBright]

Replaces http://dlang.org/phobos-prerelease/std_string.html#.capitalize with a range version analogous to http://dlang.org/phobos-prerelease/std_uni.html#.toLowerCase

[burner]

this looks redundant. first you check if it is some char than you check if the size equals dchar. why not is(ElementEncodingType!Range == dchar)

[WalterBright]

Done this way at Andrei's suggestion because it needs to allow enums whose base type is dchar, but not char or wchar which are implicitly convertible to dchar.

Note that the bulk of this code is copy/pasta from toCase.

[burner]

an string of enums?

if it copy/paste, shouldn't they be joined into a common base?

[WalterBright]

I couldn't figure out a clean way to do it. If you can, a followon PR would be great.

[burner]

I will take a look, only currently I'm starting to work on an std.xml replacement and before that I want #2995 to get merged.

[MetaLang]

toCapitalize is a poor name, but I can't think of anything much better than toCapital. Couldn't this be added as an overload of capitalize?

[WalterBright]

It won't work as an overload of capitalize(), because both accept strings.

[WalterBright]

Fixed so it has 100% unittest coverage.

[JakobOvrum]

We have join/joiner, split/splitter etc. when we have both eager and lazy versions of an algorithm. Would this convention be applicable here? capitalizer? toCapitalized?

[JakobOvrum]

toCapitalize looks like it should also be @nogc if used with representation, byChar (or is it byCodeUnit? std.utf is a mess) or handle; the original capitalize can't be tested to be @nogc-ready because of its return value, but this algorithm definitely should be.

edit:

Not suggesting that the example test should be used to verify it, but it should be verified somewhere.

[WalterBright]

the unit test allocates

[JakobOvrum]

Yes, and we need to have one that doesn't.

[MartinNowak]

Yeah, capitalize/capitalizer makes sense to me, we should strive for consistency here.

[WalterBright]

There is no consistency, and it's a little late to impute one. toCapitalize was named in the same spirit as its brethren toUpperCase and toLowerCase.

[JakobOvrum]

toCapitalize was named in the same spirit as its brethren toUpperCase and toLowerCase.

Which were also names going against Phobos precedent. There is only a little consistency, but at least it is there - let's not create yet another naming convention just for the heck of it.

[MartinNowak]

Ah OK, toCapitalizer is only the private impl, toCapitalize works for me.

[JakobOvrum]

I would have preferred capitalizer, lowerCaser, upperCaser... but it's just a consistency issue, I agree these names are perfectly apt in a vacuum.

[MetaLang]

toCapitalize doesn't make any sense. It's not proper English. You don't change a letter "to capitalize", you change it "to capital" or you "capitalize" it.

[andralex]

toCapitalize is not a phrase and I think it needs to be changed before this gets pulled

[WalterBright]

Bill wants to capitalize a sentence.

[andralex]

Yeah "to" has many meanings which actually makes it a worse choice here - you want the preposition meaning transformation akin to "into", not the function word accompanying an infinitive.

Good choices: toCapitals(), capitalizer(), toCapitalLetters()
Worse choices: crap(), whocares(), meh()

[WalterBright]

toUpperCase() - Bill wants to upper case a string.
toLowerCase() - Bob wants to lower case a string.
toCapitalize() - Fred wants to capitalize a string.

[JakobOvrum]

This also doesn't have a @nogc unittest yet.

[MetaLang]

In "a sentence".toCapitalize, Fred is not the subject. There is no Fred. A rough English translation would be (Convert) the characters "a sentence" to capitalize which obviously doesn't make any sense. It's imperative form, where the subject is implicitly "the computer". Please change it.

[andralex]

toUpperCase() - Bill wants to upper case a string.
toLowerCase() - Bob wants to lower case a string.

I think you are wrong about the use of "to" here and in fact it's the first time I'm seeing this interpretation. It's:

toUpperCase() - Convert this string to uppercase.
toLowerCase() - Convert this string to lowercase.

[MartinNowak]

Let's settle on toCapitals then.

[WalterBright]

1. changed to toCapitalized()
2. added nothrow @nogc unittest

[MetaLang]

LGTM

[MartinNowak]

Auto-merge toggled on