Re: Weird 8859-1 charset

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Sat, 15 Aug 2026 09:58:16 +0000 (UTC)
    From:        RVP <rvp%SDF.ORG@localhost>
    Message-ID:  <3490bef5-a414-1c73-aeff-2cf2c090ebeb%SDF.ORG@localhost>

  | I don't know why that is,

Pity, none of the people who have replied (for which my appreciation)
seem to have any idea.   I certainly don't want to become NetBSD's
locale expert... (no-one could be a worse choice).

  | but, this, OTOH, is clearly some typo:

  | src/share/locale/ctype/kk_KZ.PT154.src:
  |       26 MAPLOWER        <'A' - 'Z' : 'a'>
  |       27 MAPUPPER        <'A' - 'Z' : 'A'>

Yes, that is obviously broken, and easy to fix - but much of the
rest of what is in that file looks to be nonsense as well, and
it isn't really worth fixing that, and leaving the rest of it.

For example:

	MAPLOWER        <0x86  0x99>

From what I can tell (just from a web search), 0x86 is a dagger (U2020), and
0x99 is the trade mark sign (U2122).   Why exactly dagger would be considered
an upper case trade mark sign I have no idea, nor why either of them are ALPHA.

There are several like that in the list I looked at - but I might be
looking at a different KZ encoding than the one there, so I am not
intending to change anything.   We need someone with some idea what is
going on to fix it properly.

  | src/share/locale/ctype/zh_CN.GB18030.src:
  |       40 MAPLOWER        < 0x61 - 0x7a : 0x41 >
  |       41 MAPUPPER        < 0x41 - 0x5a : 0x61 >

Given that, and my even greater lack of understanding of Chinese
code sets, I didn't even look at that one.


  | > First, the Latin-1 file also contains:

  | I *think* that's used when transcoding from ISO-8859-1 to multibyte
  | ISO-2022 (CN, JP, KR):

Thanks, that's plausible.

kre