suggested changes for the collator based sort

[splitbrain]

second try, after I messed up the branch for #1

[moisesbr-dw]

Well done! Only some missing details...

[moisesbr-dw]

Excellent job! Please keep the comments as indicated, they will guide anyone interested in providing new examples in other languages.

[moisesbr-dw]

@splitbrain thanks for your suggestions. All in all, everything is OK except minor details. The most important thing missing is to substitute Sort::xxx for intl_xxx in files outside the inc folder.

I have revised everything. This way, if you are busy with other things, I can accept the changes and make the appointed corrections later.

By the way, it would be interesting to use the intl extension in other contexts -- date format, for example.

[splitbrain]

If you have all the changes ready to be applied, please go ahead. Thanks for double checking my stuff. Once this is done, we can merge it into code.

[moisesbr-dw]

@splitbrain I think you'll like to know about the Co-official statuses of German in Brazil. Just found that on Wikipedia.

[moisesbr-dw]

@splitbrain I tried the collation "an äh b c d e f g h i j k l m n os öl p q r so ßa t um üb v w x y z" but the tests failed for the umlaut pairs.
It seems that a = ä, o = ö, u = ü and s < ß in the German collation. Is it true?

[moisesbr-dw]

It seems that a = ä, o = ö, u = ü and s < ß in the German collation. Is it true?

I have commented the test code so it is utterly clear:

        // German
        // vowels with umlaut come after (thus "a" < "ä"), but are equivalent
        // in collation (thus "äpfel" < "arzt"), so a 2nd letter breaks the test
        'de' => 'a ä b c d e f g h i j k l m n o ö p q r so ßa t u ü v w x y z',

[splitbrain]

https://de.wikipedia.org/wiki/Deutsches_Alphabet#Alphabetische_Sortierung has some Details (I didn't know about): According to DIN 5007:1991, there are two variants:

1. Dictionary-Sorting (used for common words): Ä,Ö,Ü are treated exactly as A,O,U. ß is treated as ss
2. Phonbook-Sorting (used for people's names): Ä, Ö, Ü are treated as Ae, Oe. Ue

I guess variant 1 makes sense for us. In general Germans will not care at all and are fine even with the words sorted at the end.

[moisesbr-dw]

https://de.wikipedia.org/wiki/Deutsches_Alphabet#Alphabetische_Sortierung has some Details (I didn't know about): According to DIN 5007:1991, there are two variants:

English Wikipedia says the same.

ß is treated as ss

Yes, "so ßa" works, "sz ßa" fails. I'll update the comment in the test code.

I guess variant 1 makes sense for us.

The collator uses this variant. Actually, the default collator is used for German.

In general Germans will not care at all and are fine even with the words sorted at the end.

Yeah, but Spaniards would cry if their beloved "ñ" is treated like a normal "n" or put after "z" :-)

[moisesbr-dw]

OK, I've updated the comment and the collation.

        // German
        // vowels with umlaut come after (thus "a" < "ä"), but are equivalent
        // in collation (thus "äpfel" < "arzt"), so a 2nd letter breaks the test;
        // "ß" comes after "s", but is equivalent to "ss" in collation
        'de' => 'a ä b c d e f g h i j k l m n o ö p q r sr ß st t u ü v w x y z',