Windows Emacs and screen readers

Windows Emacs and screen readers

[Nikhil Nair]

[I appreciate help-emacs-windows might have been the more proper place for
this question; however, the archives indicate that previous discussion on
the topic took place here.]

Hi,

I'm blind, and want to use emacs under Windows XP.  I've tried different
versions (NT Emacs 21.3 both GUI and under cmd.exe, and Cygwin emacs 21.2,
which runs in console mode).

With the two console-mode versions, I had some terminal emulation issues
which stopped me using them fully, but those aren't emacs issues so I'll
leave them for now.

In the GUI version, I couldn't get JAWS (the screen reader I use - version
6.10) to recognise the cursor.  I'd been told that JAWS needs a blinking
bar cursor, so used blinking-cursor.el from wonderworks.com together with
a setting for a bar cursor - but no luck.  From a glance at w32term.c, I
got the impression that emacs draws the cursor graphically, but doesn't
use an ANSI character to represent the cursor, which I believe is what
JAWS needs.

However, I suspect I was reinventing the wheel, as I found some posts on
this list from 3 years ago which talked about the use of the system caret,
and JAWS working well with the GUI version; Ben Key of Freedom Scientific
(who develop JAWS) seemed quite happy with the results.  What I couldn't
tell was whether this behaviour was in version 21.3, or whether I'd need a
later development version; or whether I need to put something specific in
my .emacs file (or on the command-line) in order to get this accessible
behaviour.

Could anyone give me a pointer or two?

Cheers,

Nikhil.

Windows Emacs and screen readers

[Nick Roberts]

Nikhil Nair writes:
 ...
 > I'm blind, and want to use emacs under Windows XP.  I've tried different
 > versions (NT Emacs 21.3 both GUI and under cmd.exe, and Cygwin emacs 21.2,
 > which runs in console mode).

Perhaps Emacspeak is what you want:

Email:  raman@cs.cornell.edu
WWW:    http://emacspeak.sf.net/raman/
AIM:    TVRaman
PGP:    http://emacspeak.sf.net/raman/raman-almaden.asc
IRC:    irc://irc.gnu.org/emacspeak

Even if it isn't exactly what you are looking for, I think T. V. Raman might
be able to answer your questions.

Nick

Re: Windows Emacs and screen readers

[Jason Rumney]

Nikhil Nair <nnair@pobox.com> writes:

> However, I suspect I was reinventing the wheel, as I found some posts on
> this list from 3 years ago which talked about the use of the system caret,
> and JAWS working well with the GUI version; Ben Key of Freedom Scientific
> (who develop JAWS) seemed quite happy with the results.  What I couldn't
> tell was whether this behaviour was in version 21.3, or whether I'd need a
> later development version;

You need the later development version. Although the first changes to
use the system caret were made in 2001, a few months before 21.1 was
released, it was too close to the release to make it into 21.1, and
subsequent releases have all been bugfix releases.

Re: Windows Emacs and screen readers

[Emilio Lopes]

Jason Rumney <jasonr <at> gnu.org> writes:

> 
> Nikhil Nair <nnair <at> pobox.com> writes:
> 
> > However, I suspect I was reinventing the wheel, as I found some posts on
> > this list from 3 years ago which talked about the use of the system caret,
> > and JAWS working well with the GUI version; Ben Key of Freedom Scientific
> > (who develop JAWS) seemed quite happy with the results.  What I couldn't
> > tell was whether this behaviour was in version 21.3, or whether I'd need a
> > later development version;
> 
> You need the later development version. Although the first changes to
> use the system caret were made in 2001, a few months before 21.1 was
> released, it was too close to the release to make it into 21.1, and
> subsequent releases have all been bugfix releases.
> 

Here is the documentation of the relevant user option:

   w32-use-visible-system-caret's value is nil

   Flag to make the system caret visible.
   When this is non-nil, Emacs will indicate the position of point by
   using the system caret instead of drawing its own cursor.  Some screen
   reader software does not track the system cursor properly when it is
   invisible, and gets confused by Emacs drawing its own cursor, so this
   variable is initialized to t when Emacs detects that screen reader
   software is running as it starts up.

   When this variable is set, other variables affecting the appearance of
   the cursor have no effect.

   Defined in `C source code'.

Re: Windows Emacs and screen readers

[Nikhil Nair]

On Wed, 29 Jun 2005, Emilio Lopes wrote:

> Jason Rumney <jasonr <at> gnu.org> writes:
>
> > [...]
> > You need the later development version. Although the first changes to
> > use the system caret were made in 2001, a few months before 21.1 was
> > released, it was too close to the release to make it into 21.1, and
> > subsequent releases have all been bugfix releases.
> >
>
> Here is the documentation of the relevant user option:
>
>    w32-use-visible-system-caret's value is nil
>
>    Flag to make the system caret visible.
>    When this is non-nil, Emacs will indicate the position of point by
>    using the system caret instead of drawing its own cursor.  Some screen
>    reader software does not track the system cursor properly when it is
>    invisible, and gets confused by Emacs drawing its own cursor, so this
>    variable is initialized to t when Emacs detects that screen reader
>    software is running as it starts up.
>
>    When this variable is set, other variables affecting the appearance of
>    the cursor have no effect.
>
>    Defined in `C source code'.

Thanks to both of you for these helpful comments, and to Lennart Borgman
who pointed me to the precompiled version at
http://ourcomments.org/Emacs/EmacsW32.html .

I've just tried to install the development version I found there, on XP
SP2 (non-admin account), keeping all of the defaults except the
installation directory, which I put in my user account's filespace.  It
seemed to install, but then wouldn't run; I get:

---------------------------
C:\Documents and Settings\nnair\Emacs\setup\bin\gnuclientw.exe
---------------------------
Could not start runemacs --load="C:/Documents and Settings/nnair/Emacs/setup/bin/gnuservauto.el"
---------------------------
OK
---------------------------

(the above is the result of cutting from a message box with C-c.)

Any hints appreciated.  I did the installation from an unprivileged
account, but since the option it gave me was to do a full installation
except anything requiring privileged access, I presumed this would be
fine.

Cheers,

Nikhil.

NTEmacs documentation (was: Windows Emacs and screen readers)

[Eli Zaretskii]

> From: Emilio Lopes <eclig@gmx.net>
> Date: Wed, 29 Jun 2005 07:30:27 +0000 (UTC)
> 
> Here is the documentation of the relevant user option:
> 
>    w32-use-visible-system-caret's value is nil
> 
>    Flag to make the system caret visible.

Which reminds me: would someone wish to volunteer to write an appendix
for the manual that documents MS-Windows specific features and
misfeatures?  Currently, the only Windows aspects that are documented
(and buried in the appendix dedicated to the MS-DOS port) are
subprocesses, text vs binary files, and the w32-pass-alt-to-system
variable.  That leaves a lot to be desired, and is a shame, I think.

Any takers?

Screen reader detection broken [Was: Re: Windows Emacs and screen readers]

[Nikhil Nair]

On Wed, 29 Jun 2005, Emilio Lopes wrote:

> [...]
> Here is the documentation of the relevant user option:
>
>    w32-use-visible-system-caret's value is nil
>
>    Flag to make the system caret visible.
>    When this is non-nil, Emacs will indicate the position of point by
>    using the system caret instead of drawing its own cursor.  Some screen
>    reader software does not track the system cursor properly when it is
>    invisible, and gets confused by Emacs drawing its own cursor, so this
>    variable is initialized to t when Emacs detects that screen reader
>    software is running as it starts up.
>
>    When this variable is set, other variables affecting the appearance of
>    the cursor have no effect.
>
>    Defined in `C source code'.

Hi,

I've now had a chance to play with Windows emacs, and, with this feature
enabled, it works very nicely with JAWS (a popular screen reader).
However, if it's supposed to be autodetecting screen readers, this isn't
working at all: I had to add (setq w32-use-visible-system-caret t) to my
.emacs.

Although I don't really use it any more, I have a version of Window-Eyes
(another popular screen reader) from a couple of years ago (as opposed to
my version of JAWS, which is up to date).  Emacs wasn't able to autodetect
this, either.

I hope it might be possible to fix this before it's released...

Cheers,

Nikhil.

Re: Screen reader detection broken [Was: Re: Windows Emacs and screen readers]

[Eli Zaretskii]

> Date: Fri, 1 Jul 2005 17:50:54 +0100 (BST)
> From: Nikhil Nair <nnair@pobox.com>
> 
> However, if it's supposed to be autodetecting screen readers, this isn't
> working at all: I had to add (setq w32-use-visible-system-caret t) to my
> .emacs.
> 
> Although I don't really use it any more, I have a version of Window-Eyes
> (another popular screen reader) from a couple of years ago (as opposed to
> my version of JAWS, which is up to date).  Emacs wasn't able to autodetect
> this, either.

If you tell us what are the possible ways to autodetect screen
readers, it's possible that we will teach Emacs to turn that option on
automatically when a screen reader is active.

Please note that to do the right thing, it is not enough to detect
that a screen reader is installed: Emacs should find out whether it is
actually active for the current user.

TIA

Re: Screen reader detection broken

[Jason Rumney]

Nikhil Nair <nnair@pobox.com> writes:

> I've now had a chance to play with Windows emacs, and, with this feature
> enabled, it works very nicely with JAWS (a popular screen reader).
> However, if it's supposed to be autodetecting screen readers, this isn't
> working at all: I had to add (setq w32-use-visible-system-caret t) to my
> .emacs.

Thank you for reporting this. It turns out the detection was being
done in the wrong place, so was happening when Emacs was dumped, not
every time it started. I have fixed it now.