NLS files and FONTSIGNATURE

[Pablo Saratxaga]

Kaixo,

I'm planning to create/fix/check all the possible NLS files.

There is, however, an entry on those files that is problematic for me,
namely the LOCALE_FONTSIGNATURE one.

>From the Microsoft docs I understand it is made of 3 bitfields,
one 128bit for unicode  coverage, and two 64bit ones for codepage
("default" and "supported").

For those last two there is no problem, after the normal
endianess switching I get an hex number with bits that make sens.
eg for rus.nls file I have a fontsignature being:
 L"\x0203\x8000\x3848\x0000\x0000\x0000\x0000\x0000\x0004\x0000\x0000\x0002\x0004\x0000\x0000\x0202"

the default and supported codepage parts are respectively:

 \x0004\x0000\x0000\x0002
 \x0004\x0000\x0000\x0202

after switching \x(1)\x(2)\x(3)\x(4) to 0x(4)(3)(2)(1) I get:

0x0002000000000004 and 0x0202000000000004

which perfectly match bits 2, 49, 57; code pages 1250, 866, 855

But for the unicode coverage part I'm lost...

In the LOCALE_FONTSIGNATURE it is:
 \x00a3\x8000\x7878\x38c9\x0016\x0000\x0000\x0000

In order to get a 128 bit number that makes sense I had to do
the following switching: 

 \x(1)\x(2)\x(3)\x(4)\x(5)\x(6)\x(7)\x(8) to 0x(2)(7)(6)(5)(4)(3)(8)(1)

but it seems strange to me.

Also, font signatures for other nls don't make sense with that substitution.

So; how should I made the conversion between the format used in *.nls
files and the 128 bit number used as bitfield to calculate the flags?
Are the values in *.nls files trusty?

Thanks

-- 
Ki ça vos våye bén,
Pablo Saratxaga

http://chanae.walon.org/pablo/		PGP Key available, key ID: 0xD9B85466