ispell mode
* installation
Simply put it all in your load path, add
require ("ispell_init");
to your .jedrc, and any keybindings you want.  And set up your dictionaries.
* dictionaries
On Debian the dictionaries should be automatically initialised by the
dictionaries-common package.  On other systems, use
ispell_add_dictionary(), aspell_add_dictionary() or hunspell_add_dictionary().
  
hash_name: the name by which ispell knows the language (a .hash file in
  /usr/lib/ispell)

letters: letters that constitute words.  For latin1 languages, the default
  should be all right.

otherchars: chars that may appear inside words, or outside words.  In
  English it's "'", in other languages it may be "-'".  Old versions of
  dictionaries_common don't provide this argument, so you have to update.

ext_char: something that's put in the ispell command line after a "-T" flag
  and should match a stringtype defined in /usr/lib/ispell/(language).aff.
  So it's not the same as Emacs ispell's extended-character-mode or
  character-set.  If you're European try latin1.  Dutch doesn't seem to
  need it but in German an 'ä' will be turned into an '"a' without it.
  This option is only for ispell_add_dictionary(), not for
  aspell_add_dictionary().

options: any other args you want to pass.

wordlist: just needed for ispell_complete, it would be annoying to have to
  set the language for ispell AND a wordlist for completions.  For Debian
  users, this one is not automatically filled in.

* ispell and utf-8
For checking utf-8, it's recommended to use aspell.  Ispell.sl does not work
with ispell in utf-8 mode, though flyspell.sl should.
* using (i|fly)spell with other modes
** checking TeX
When in TeX mode, ispell_region() will tell the ispell process to switch
to TeX mode.  Ispell will not check the markup.  However, when a word is
misspelled, the rest of the line is again sent to ispell like this:

check out this fomrula $ A^2 +
B^2 = C^2 $

First it will stop at fomrula, then "formula $ A^2 +" will be sent
again to re-check the correction and check the rest of the line (much
easier to program that way).  The problem is that in TeX, "$" starts
and ends a math environment which ispell does not check, and ispell
does not see its input as unrelated lines, but tries to keep track of
the environment.  The first "$" is seen twice, so the third "$" will
start a new math environment for ispell, and the rest of the buffer
will not be checked.  I don't know how much of a problem this is, since
I don't use TeX.  BTW I read somewhere that the `$' to delimit math
environments is deprecated.
*** flyspelling TeX
To flyspell TeX while keeping your syntax highlighting use a hook
like this:

define latex_mode_hook()
{
  define_blocal_var("flyspell_syntax_table", "TeX-Mode");	
}
Flyspell doesn't always work with DFA, so it turns DFA mode off when a
blocal flyspell_syntax_table is set.
** checking html
Doesn't work.  Since ispell does not have an interactive command for
turning html parsing on, you would have to create a dictionary entry
for html:

ispell_add_dictionary("html", "british", NULL, "-'",
"/home/paul/htmlwords.txt", " -h");

Htmlwords.txt would be a list of html keywords for word-completion.

You can set this in the html mode hook:
define html_mode_hook()
{
 ispell_change_local_dictionary("html");
 define_blocal_var("flyspell_syntax_table", "html");
}

This could be expected to have the same shortcomings as [checking TeX].  If
you'd check

get more kool jedmodez from <a
href="http://jedmodes.sf.net">

you could expect it to complain about "kool" but skip over "jedmodez" since
it would think this is part of a HTML tag.  As it turns out, html parsing
works differently between a full-screen ispell and ispell -a.  The
full-screen ispell deals correctly with tags that span lines, but ispell -a
will echo the lines between the opening and closing bracket without an empty
line to indicate it has finished processing the line.  Ispell.sl does not
know how to deal with this and crashes.
** checking email
Email.sl has an ispell_message() function that will check the body of of
the message, and an ispell_region_hook that will skip past quoted text.
Get it from http://jedmodes.sf.net.
* FAQ
** Undefined Name: ispell_region: Function did not autoload
You installed ispell.sl in jed's lib/ directory and didn't remove
ispell.slc. Put non-standard modes in their own directory.
** Invalid UTF-8 encoded string
You're running in utf-8 mode, and the 'letters' and/or 'otherchars'
arguments of your ispell_add_dictionary or aspell_add_dictionary calls are
not encoded in utf-8.  Use something like
if (_slang_utf8_ok)
{
  % utf-8 settings
}
else
{
  % latin1 settings
}
* license
This package is GPL'ed.  No warranty.
You can get the latest version at http://jedmodes.sf.net/mode/ispell
* thanks
Kudos to Günter Milde, Rafael Laboissiere, Agustin Martin and JED.