From mboxrd@z Thu Jan  1 00:00:00 1970
From: erik quanstrom <quanstro@quanstro.net>
Date: Mon,  5 Jan 2015 14:05:17 -0800
To: 9fans@9fans.net
Message-ID: <2926311e74db0ddc2310008906226507@lilly.quanstro.net>
In-Reply-To: <5e85d81160785cdf717f01a8c0649731@quintile.net>
References: <5e85d81160785cdf717f01a8c0649731@quintile.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Subject: Re: [9fans] I don't understand utf8 (it seems)
Topicbox-Message-UUID: 39531eaa-ead9-11e9-9d60-3106f5b1d025

On Mon Jan  5 13:48:47 PST 2015, steve@quintile.net wrote:
> I am trying to parse a stream from a tcp connection.
>=20
> I think the data is utf8, here is a sample
>=20
> 	 20 2d 20 c8 65 73 6b fd 20 72 6f 7a 68 6c 61 73
>=20
> which when I print it I get:
>=20
>      -       e  s  k       r  o  z  h  l  a  s          =20
>            ^          ^
>         missing    missing
>=20
> there are two missing characters. Ok, bad UTF8 perhaps?
> but when I try unicode(1) I see:
>=20
> 	unicode c8 fd
> 	=C3=88
> 	=C3=BD
>=20
> Is this 8 bit runes? (!)
> Is there a name for such a thing?
> Is this common?
> Is it just MS code pages but the >0x7f values happen (designed to) to m=
ap onto the same letters as utf8?

latin1 has this property that if you embed the byte in a rune-sized chunk=
, then it's
a valid Rune.  but latin1 is invalid utf-8.

  the reason that unicode(1) failed to meet expectations, is that the des=
ire was=20
to convert the supposed utf-8 0xc8 to a codepoint, but what unicode did w=
as
convert the codepoint 0xc8 into utf-8.

- erik