Old Icelandic sort order: patch for ctype-uca.c

[Sven Müchert]

Hi!

I'm new here, so hi! :-)

I noticed that 'COLLATE utf8_icelandic_ci' does not account for Old Icelandic. The sort order for the additional characters of Old Icelandic should be:

a..d,ð,e..z,þ,æ,œ,ǫ,ø

Characters with macron e.g. ǭ, should probably be treated as if they had an acute accent, so they are different chars order after the unaccented character. The problem chars are œ and ǫ, which are currently treated like 'o' instead of sorted by the above order. The others are in correct order already by the Modern Icelandic order.

Sometimes, ę is used, and should be secondary after æ.

I searched the source code for the corresponding table and found it in strings/ctype-uca.c. There is a table 'icelandic' that I modified to read as follows:

static const char icelandic[]=
"& A < \\u00E1 <<< \\u00C1 << \\u0101 <<< \\u0100 "
"& D < \\u00F0 <<< \\u00D0 "
"& E << \\u0119 <<< \\u0118 " /* e << e ogonek */
"< \\u00E9 <<< \\u00C9 << \\u0113 <<< \\u0112 " /* < e acute << e macron */
"& I < \\u00ED <<< \\u00CD << \\u012B <<< \\u012A "
"& O < \\u00F3 <<< \\u00D3 << \\u014D <<< \\u014C "
"& U < \\u00FA <<< \\u00DA << \\u016B <<< \\u016A "
"& Y < \\u00FD <<< \\u00DD << \\u0233 <<< \\u0232 "
"& Z < \\u00FE <<< \\u00DE " /* thorn */
"< \\u00E6 <<< \\u00C6 " /* ae */
/* "<< \\u00E4\\u0301 <<< \\u00C4\\u0301 " e ogonek w/ acute: HOWTO? */
/* "<< \\u00E4\\u0304 <<< \\u00C4\\u0304 " e ogonek w/ macron: HOWTO? */
"<< \\u00E4 <<< \\u00C4 " /* a diaeresis */
"< \\u0153 <<< \\u0152 " /* oe */
"< \\u01EB <<< \\u01EA " /* o ogonek */
/* "< \\u01EB\\u0301 <<< \\u01EA\\u0301 " o ogonek + acute: HOWTO?
"<< \\u01ED <<< \\u01EC " o ogonek w/ macron */
"< \\u01ED <<< \\u01EC " /* o ogonek w/ macron */
"< \\u00F6 <<< \\u00D6 << \\u00F8 <<< \\u00D8 " /* o slash = o diaeresis */
"< \\u00E5 <<< \\u00C5 "; /* a ring */

This does not alter the original sort order but only add support for Old Icelandic. I test-compiled it and tested mysqld for correct order. One problem remains: I don't know how to add sort order for composite chars (ę with macron or acute and ǫ with acute). If I put \\u0301 or \\u304 behind the base chars, the table is rejected.

Anyway, the above is an improvement, I think.

I hope this is right place to post this patch?

Sven