[orca-list] buffering

[Daniel Dalton <d dalton iinet net au>]

Hi,

I'm sure this topic has been discussed before. I was in discussion with
a couple of people from another mailing list, had to use jaws for a day,
(and hated it), but we got onto the topic of virtual buffering. I don't
think it's the best idea myself, but I think it is good at times. So,
would a key to "buffer" the current page, help at all with:
- speed of navigating pages when blind?
- Responsiness 
- Accessibility.

Of course, it is hardly necessary, but could be useful at times. I'm
just curious has it been considered? I'll paste some of the emails
below, to see what people think.

Daniel.

Start debate :)

----- Forwarded message from Kerry Hoath <kerry gotss net> -----

From: 
To: Daniel Dalton <d dalton iinet net au>
Cc: vip-l <vip-l softspeak com au>
Subject: Re: VIP-L: Jaws
Date: Fri, 27 Mar 2009 20:41:05 +0900
X-Mailer: Microsoft Outlook Express 6.00.2900.5512

Those who don't want to read techy comparisons, press delete or control-d  
now or whatever the equivalent delete message method is in your chosen  
operating system happens to be.

Whilst I read this message with interest 2 things occur to me.
1: this message is targetted at a technical audience. It is probably better 
targetted to the Jaws/Orca list.
Hearing about what you apparently don't like about Jaws but love about 
Linux and lynx probably isn't going to make lots of people switch to Linux  
overnight, and the content of these messages is too technical for many 
users to grock.

People who do want to try Linux can certainly obtain Ubuntu, vinux
http://www.vinux.org.uk
or arch linux talking install media.
Vinux might be a good choice for many as it comes up talking, has support  
for many braille displays, (automatic support for many USB ones) and will  
provide full-screen magnification with the press of a key sequence.

You have been brought up with Linux or chosen that OS and believe that 
their way is the best way, or at least for you.
There's no point knocking the other ways until you have thoroughly  
experienced them and their strengths and limitations.

Many others may choose to stick with Windows, as that is what they are  
comfortable with. That is their choice.

2: This message seems to talk about the way you expect and want things to  
work, and what you think is better.
Whilst you may feel you have some good ideas in this field, i'm sure  
Freedomscientific also have ideas that have and do work for them and have  
done for aproximately 10 years now.

Windoweyes also uses a virtual buffer, although they call it browse mode. 
It appears that 2 separate screen reader companies find value in a virtual  
buffer for windows and Firefox.
NVDA virtualizes pages also to my knowledge.
This method is affective, fast and does work for a lot of people and  
continues to do so. If the method had been floored surely it would have 
been disguarded years ago.


Perhaps if you want your Orca and your object rendered page then that is 
the solution _you_ should be using.
For the majority of us, the virtual buffer, forms mode, quick keys and the  
like work well for rich media content pages that contain flash and web 2.0  
elements under Windows.
Try your solution on facebook and other web 2.0 pages and see how it stacks 
up, perhaps you will gain an understanding for how Windows does things in a 
different but valid way.

Also why not try Jaws out and learn how it works rather than plying the 
list with questions? You'll have a far better understanding of how the 
underlying system works from a user interface perspective.

The reason why Orca can interact with firefox as it does is that it hooks  
into the underlying API of the browser and understands how objects render  
and what they are.
This code and methods will require constant evolution as firefox and gnome  
mature.

Are you sure that what you hear aproximates the page layout? How do you 
know how wide a table column is? How do you know the spacing between 
graphics or text? In my experience Orca and firefox aproximate as well.
As previously noted, Orca only works with Firefox to my knowledge, Jaws's  
virtual buffer works with Internet explorer, Firefox, and other rich edit  
controls providing a uniform method of accessing these across multiple  
applications.


If you want a different take on things try NVDA and firefox 3 under 
Windows; although I think as stated above NVDA whilst hooking into the 
internals of firefox virtualizes.

Also note that lynx destroys the formatting of a page to display it on a  
fixed-width terminal or braille display.
Most text unless specifically formatted is in a proportional font, which is 
to say that an I is narrower than an M.
Courier and times are fonts that do have all letters the same width.
In a text display, an I is as wide as an M, (or takes up the same 
horizontal space) and you allways get the same number of characters per 
line.

Many fonts in a GUI will give you more narrower letters so you'll get more 
I characters on a line than you will ms.
did you know this? did Orca expose this to you or lynx?
Ever wondered why tables in html format so badly on a fixed-width terminal? 
There are some of your reasons why output manipulation is required.

You have not had much experience with Jaws's virtual buffer and therefore  
have no idea whether or not it is faster or slower to use than your  
solution.
Initially it will seem slower as you are not used to the methods or  
interfaces but as you learn to use them you can make an informed  decision  
as for what is best for you.

I personally own a Mac, and do not enjoy the way it does things and the  
speed at which voiceover operates. It is my choice not to use voiceover, 
and my reasons are gleaned from my experience with the operating system and 
environment.
My objections to voiceover are made based on technical experience, and not 
a comparison to how another operating system does things.

Once you have experience of the ccompetition you will be able to make your  
own decisions on what you prefer and choose to use.
You may find however as your education continues and you get into  
potentially TAFE or University you will have to use webCT or Blackboard,  
systems that do require a screen reading solution with the ability to  
present rich multi-media content in a way vision impaired can understand. 
It will be interesting to see which of the many solutions suits you in 
these environments.
I know plenty of people who used to use Linux, but ended up using a mix of  
Windows and Linux for various reasons such as
the need or desire to access an outlook mail system, access to pdf  
documents, the ability to edit complex office documents and write macros 
and similar.
the solutions you use now, did not exist 3 years ago, in their current 
form, (firefox and Orca) and what you'll use in 3 years time may not be 
like what you use today.

I use Windows for a lot of things for the above reasons, including sound  
editing, virtual machines in vmware work station with snapshots etc and for 
the web 2.0 experience.
I prefer Jaws to Windoweyes, but I can and have used both.
It is up to you to settle on what suits you best to achieve the work you  
wish to do.
Computers and software are tools to be applied by a person to the day to 
day problems they face.
Regards, Kerry.


----- End forwarded message -----

I've used both orca and JAWS btw, if anyone is curious :)
[jaws for over 3 years]

----- Forwarded message from Daniel Dalton <d dalton iinet net au> -----

Date: Fri, 27 Mar 2009 18:33:43 +1100
From: Daniel Dalton <d dalton iinet net au>
To: vip-l softspeak com au
Subject: Re: VIP-L: Jaws

On Tue, Mar 24, 2009 at 06:01:28PM +1000, Aaron Howell wrote:
> Hi,

Hi,

Please see below

> > Can't you use other reading commands? Like read column row, word,
> > sentence etc?
> > Perhaps it helps I use braille.
> Yes and no. That works ok for simple pages, but in today's world of web 2.0/ajax/flash,  based pages,
> having a virtual buffer  is very handy because it allows the screen reader to present a static 
> representation  of the page contents,
> without issues like onmouseovers being activated while you are trying to read.

Wouldn't having a key to "freeze" or "buffer" the page then be best, not
default to buffering?

> I also think that interacting with Flash would be much harder (if its even possible at all) without the use 
> of a virtual page representation, does orca do that?

Dunno, as I said I mainly use lynx, but I'll play with it. Do you mind
if I send this to the orca mailing list (I'll remove all email
addresses), to get their opinion. I think a key to buffer the page could
be handy, but I still think buffering the page straight away is pretty
dumb... :-)

> > > reformat the page in a way that makes a lot more sense. Jaws has 2 ways
> > > of doing this but the default way works for most people. Usually each
> >
> > How can someone then understand how the page actually looks? eg. a web
> > designer?
> Possibly through use of the Jaws cursor,
> though from the perspective of someone who has no sight, even that will only take you so far.
> Regardless of whether you use a virtual buffer to render the page or you rely on trying to present it in an 
> object oriented manner as screen readers like nvda, orca, and VoiceOver on the Mac do, you still only ever 
> have an approximation of where things are located on the page, and unless you actively seak out the 
> information, no idea of how colours, graphics, etc, are positioned. If I was designing anything other than 
> the most rudamentary webpage I would be looking for sighted assistance.

True, but I guess it gives you a better idea...

> > > link is on a separate line. Things like using tables for setting out
> >
> > Why each link on a separate line?
> Ease of navigation. You can hit enter when you hear a link read, and you will be absolutely sure you are 
> clicking on the link you intended.
> If you render more than one link per line then you have to make sure that the mouse pointer is physically 
> located over the correct one. This is ok if you are tabbing around, but if you're arrowing up and down it 
> isn't.

Don't you have keys to jump from link to link? We do in orca, and also
in lynx! Out of curiosity with JAWS, can you jump to say the 12th link
on the page? I can do this with lynx/brltty, but not sure about firefox.

> > > you can easily jump between frames. You can instantly jump to non linked
> > > text. There are loads of other advantages you'd only really see if you
> > > played around with it a bit. You can also see exactly how the page
> >
> > Couldn't I do this with out a virtual buffer?
> No. The virtual buffer is what allows the screen reader to know exactly where you are on a page, as opposed 
> to where the mouse cursor happens to be pointing.

So how come orca can find tables, links forms etc?
How come I can jump to a block of text? (considering orca doesn't use
any sort of buffering)

> > Couldn't this be a problem if you are quickly browsing many pages?
> Not noticeably, unless the pages were really very sparsely designed, ie, basically text only,
> the page is usually rendered in the virtual buffer by the time all the graphics have finished loading.

Alright then, that's reasonably good. I still think having a key to
buffer the page or a configuration setting to buffer it would be best.


> > > 4. What is better about this?
> > >
> > > Far easier to navigate especially on complex pages such as
> > > www.facebook.com. Try reading that page line by line as it is and it'll
> > > make very little sense.
> >
> > Perhaps, I should really give that a go actually with orca. 
> You'll find that very frustrating I think.

I probably will :) Will give it ago at some stage and let you know how
it goes. As I said, having a key to buffer these types of pages may be
good, but then for other pages buffering gets in the way.

> > That's good then, but what about for users that can't aford this
> > update... It seems to me like a pretty basic thing almost a bug if words
> > like link are copied in my document, so I personally would expect a free
> > bug fix, although I know that wouldn't happen...
> There are cases I can see for wanting it both ways, so its not a bug but a design feature.
> The original behaviour is compatible with more applications, because it doesn't require the receiving 
> application to be html aware.
> If you are copying to a text only application like Notepad, you will still see the contents of the virtual 
> buffer rather than the actual page.
> Given that without the virtual buffer the only way to cut and paste would be with the mouse, I don't see 
> this as too much of a tradeoff.

Surely for big pages this would be annoying? I guess lynx puts link
numbers in when I dump the page, but I then have the option of using my
mouse to cut and paste or brltty... 

> > that use jaws that everything is virtual?
> > (in Internet Explorer/firefox)
> Only power users, but then, they are the only ones who really need to.
> For day to day browsing there isn't any need to know exactly where a particular element is located on a 
> page, etc,
> so virtual browsing works just fine.

I guess that's a valid point. Although I think jaws should aim to meet
power users requirements a bit better.

> You seem to do just fine with Lynx for the majority of your browsing,

Yes

> and that gives you far less than either Orca/Firefox or Jaws and its virtual buffer, so unless you have a 
> particular need for knowing exactly how things are laid out visually, then the virtual buffer approach 
> would seem to be superior.

Far less what? It puts links on the same line, but I'm not sure how much
lynx messes with the view of the page. I guess what sighted people would
see with lynx is the same as I would see on my braille display...

> If applications don't follow the api to the letter though, (or in the case of Orca, if they aren't gtk 
> based),
> then you get no access at all.

Yes, but if an at-spi setup is written for qt then that should provide
access to that tool kit...
> (quite a number of negative VoiceOver reviews have been written by people who's biggest gripe is that it 
> doesn't work the way their Windows screen reader does),

Well, things are different... :-) The windows screenreader isn't
necessarily the correct way, and the linux way isn't always the correct
way... Surely differences would be a positive? More out there, more
range to choose from?

Thanks,

Daniel.
*** VIP-L is administered by Tim Noonan tim timnoonan com au
To Leave the list, send a blank message to
vip-l-leave softspeak com au
To Join VIP-L, send a blank message to
vip-l-join softspeak com au
the VIP-L info and settings web page is at 
https://lists.pacific.net.au/mailman/listinfo/vip-l
Disclaimer: VIP-L is a free community service provided by Tim Noonan Consulting Pty Ltd www.timnoonan.com.au
While reasonable efforts are taken by the administrator to ensure that messages are accurate
and appropriate to the list's scope,
he is unable to take any responsibility for the actual content of member's posts
or for the actions of list members.

----- End forwarded message -----