Re: [orca-list] Color name support just added to Orca master

[Joanmarie Diggs <jdiggs igalia com>]

Hey Christopher.

All you have to do is convert the RGB values (which we get via AT-SPI2)
into hex and look them up. The only tricky part was finding the nearest
match when the color is not exactly one of the items in the table. I
googled and found an answer to that problem. (Problems are like plots:
There are no new ones. <smiles>) And then I tweaked that solution
because really dark and really light colors had nearest matches of black
and white respectively. The algorithm might need a bit more tweaking,
which is something we'll learn over time.

The real problem, however, is when applications (or more likely their
toolkits) fail to expose the correct RGB value to us. In the case of
Epiphany/WebKitGtk, I'm pretty sure the problem is that the colors of
parent nodes are not being taken into account when reporting colors to
us. And I have vague recollections of Alex Surkov (from Mozilla) solving
this problem for Gecko. In the case of Gtk+ apps and VTE, I'm pretty
sure there's not an inheritance/parent node problem, but a failure of
those toolkits to update the attributes exposed to us. But, again, soon
I'll look into that and file the necessary bugs so the developers of the
toolkits in question can fix them properly.

Take care.
--joanie

On 05/11/2014 10:44 AM, Christopher Chaltain wrote:
> Wow, I'm curious how you did this. I gave it some thought a few years
> ago, and I did find a table mapping RGB values to color names, but I
> couldn't find an algorithm on line and I couldn't come up with one
> myself. I didn't spend that much time on it though, and I'm not the
> smartest guy in the world, so it doesn't surprise me that someone else
> cracked this nut, but I am curious how you did it. Feel free to ignore
> this question or respond to me privately if you don't want to discuss it
> or take up any list traffic on it. I'm just curious.