From kde-core-devel  Fri Feb 18 18:00:27 2005
From: Mashrab Kuvatov <kmashrab () sat ! physik ! uni-bremen ! de>
Date: Fri, 18 Feb 2005 18:00:27 +0000
To: kde-core-devel
Subject: Re: [PATCH] KSpell Unicode problem (BR#86940)
Message-Id: <200502181900.31549.kmashrab () sat ! physik ! uni-bremen ! de>
X-MARC-Message: https://marc.info/?l=kde-core-devel&m=110874994207670
MIME-Version: 1
Content-Type: multipart/mixed; boundary="--nextPart1780598.KK6VRWRiOX"

--nextPart1780598.KK6VRWRiOX
Content-Type: text/plain;
  charset="utf-8"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

Hi Luk=C3=A1=C5=A1,

first of all thanks for looking at the issue. Waldo, thanks too.

On Friday 18 February 2005 13:29, Luk=C3=A1=C5=A1 Tinkl wrote:
> Hmmm, unfortunate situation... Mashrab's patch breaks with older aspells,=
=20
> that's what I feared. Furthermore, KSpell+ASpell5 currently can't cope wi=
th=20
> any utf8 text at all. Attached patch fixes that.

To my knowledge (correct me if I'm wrong) neither Aspel < 0.60 nor Ispell
supports utf8 spellchecking, they treat input/output in 8-bit encoding. Tha=
t's
why, actually, I was surprised to see UTF-8 in a list of encodings of contr=
ol
center. One could argue that it is possible to pass --encoding=3Dutf8 to
Aspell-0.50, but from the documentation of Aspell-0.50 [1]=20
<quote>
encoding=20
(string) The encoding the input text is in. Valid values are ``utf-8'',=20
``iso8859-*'', ``koi8-r'', ``viscii'', ``cp1252'', ``machine unsigned 16'',=
=20
``machine unsigned 32''. However, the aspell utility will currently only=20
function correctly with 8-bit encodings. I hope to provide utf-8 support in=
=20
the future.
</quote>

What spellchecker are you using? If Ispell, are you sure it is not a wrapper
around Aspell (some distros dropped Ispell)? If you do not have Aspell-0.60,
how do you spellcheck utf8 texts (like you said Aspell-0.50 cannot do that)?

Thanks for patch, I'll definitely try it once at home. However, I doubt it
solves the problem. Currently, the spellchecking i.e., pass a word to Aspell
and get suggestions, is working very well. No way to blame backend.
The problem, IMHO, is in what happens with suggested word. Namely,
there is a variable called posinline, from how I understand the code it giv=
es
the position of a word being checked in a string. Later, lastpos is
calculated to figure out which word to replace/highlight. The part of a code
which I commented out does crazy things resulting in wrong posinline.
Did anybody understand what I said? It seems it is not clear even to
myself. :-)

1.=20
http://aspell.net/0.50-doc/man-html/4_Customizing.html#SECTION0052300000000=
0000000

Cheers,
Mashrab.

PS. I faked a reply cutting&pasting from the archive, since I was not in th=
e=20
list. Now I subscribed.
=2D-=20
Mashrab Kuvatov
Ph.D student
University of Bremen, IUP
Home-page: www.sat.uni-bremen.de/members/mashrab
PGP key: www.uni-bremen.de/~kmashrab/kmashrab.asc

--nextPart1780598.KK6VRWRiOX
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQBCFi0/iSmHahjHWuoRAuL7AKCB/OaAQT7RH0ieg4KC0t1B8SI8PwCghiRM
mxEcIeq/6RtOvivZ0L1gR+s=
=yix9
-----END PGP SIGNATURE-----

--nextPart1780598.KK6VRWRiOX--