[g-a-devel][Fwd: Re: Info inquiry about at-poke]

[Michael Meeks <michael ximian com>]

For the archive ...

-- 
 mmeeks gnu org  <><, Pseudo Engineer, itinerant idiot

--- Begin Message ---

• From: Michael Meeks <michael ximian com>
• To: silvia zhao sun com
• Cc: browser-china-atf <browser-china-atf sun com>
• Subject: Re: Info inquiry about at-poke
• Date: 14 May 2002 17:42:03 +0100

Hi Silvia,

On Tue, 2002-05-14 at 11:33, Silvia Zhao wrote:
> I am from SUN China Engineering and Research Institute and be a member
> of Accessibility Task Group. Our team had planned to develop a test tool
>  to test the accessibility features of Mozilla. Fortunately, we got your
> at-poke. It will greatly save our energies.

	Wow :-) I'm glad you like it, I wrote it to be useful. There are a good
number of things that need doing to expand it's capabilities especially
report generation / automated unit testing etc.

> But this test tool seems not perfect yet.

	Indeed, it's quite rough still.

>  Are you going to continue this work? Would you like to tell us
> your plan? This information will do help us in the development. We
> appreciate your help.

	Sure - well, I'm most happy to give you pointers as to what needs
working on.

	The most irritating thing about at-poke - at the current time is the
fact that cspi/libspi provides no consistant way to synchronize the
local a11y tree copy with the remote copy.

	This is a problem since the GtkTreeView / Model expects a consistant
model - ie. it's no good re-ordering the nodes underneath the tree
without emitting signals on the model each time.

	Thus any degree of dynamism in the application - opening new windows
etc. causes at-poke problems [ to say the least ] that can be worked
around partially by collapsing and re-expanding nodes to hide the
changes.

	To fix this is a little chunk of work, and requires remembering the
parentage of every node we expose localy, since as soon as we are
notified of a node's death, it is impossible to determine where it is in
the tree :-), so that needs doing, and we could use that information to
accelerate the tree view quite considerably as well.

	I currently have no immediate plans to work on at-poke, but I'm
reviewing all patches still. Bill Haneman is also working on it - it
seems for his Java bridge test. The best place to ask about this sort of
thing is the gnome-accessibility-devel mailing list, to which I
subscribe [ but CC me if you want a quick response ].

	I'd like to forward this mail there, for the archives - may I ?

	Please do let me know what specifically you were thinking of working
on, and whether I can help you in the above task / whether you think you
can get your teeth into it or not.

	Great to find someone else working in this area,

	Warmest regards,

		Michael.


-- 
 mmeeks gnu org  <><, Pseudo Engineer, itinerant idiot

--- End Message ---