I actually left the order out of the original test intentionally. The order is completely predictable on Windows and Wine (the window docked first ends up closer to the edge), but X docks work differently. They are controlled by the window manager.
If we ever need to create real X docks for appbars (unlikely because of how rarely they're used, but possible), the order will be defined by the window manager, and it won't be possible to reliably pass the test.
On Wed, Oct 21, 2009 at 4:43 AM, Paul Vriens paul.vriens.wine@gmail.com wrote:
Hi,
"Fixes" a long standing failure on at least my Vista box. After the removal of windows[0] there are no ABS_POSCHANGED messages around like on other boxes.
Marked as broken() as there seem to be only 2 Vista boxes that have this issue. I'm running Vista Ultimate maybe that's it?
This failure has been discussed with Vincent in the past but the thought was that we were removing the wrong window. The added tests shows however that the order of the windows is correct when we start.
Changelog Fix a test failure on some Vista boxes
-- Cheers,
Paul.
On 10/21/2009 04:50 PM, Vincent Povirk wrote:
I actually left the order out of the original test intentionally. The order is completely predictable on Windows and Wine (the window docked first ends up closer to the edge), but X docks work differently. They are controlled by the window manager.
I just added that ok() message to show that on these Vista boxes the order it the same as on all other boxes and thus the test failure is because of brokenness.
So you are fine with this patch?
I would prefer that the test pass if the order turns out to be different, to leave possibilities open for future implementations.
On Wed, Oct 21, 2009 at 10:00 AM, Paul Vriens paul.vriens.wine@gmail.com wrote:
On 10/21/2009 04:50 PM, Vincent Povirk wrote:
I actually left the order out of the original test intentionally. The order is completely predictable on Windows and Wine (the window docked first ends up closer to the edge), but X docks work differently. They are controlled by the window manager.
I just added that ok() message to show that on these Vista boxes the order it the same as on all other boxes and thus the test failure is because of brokenness.
So you are fine with this patch?
-- Cheers,
Paul.
On 10/21/2009 05:42 PM, Vincent Povirk wrote:
I would prefer that the test pass if the order turns out to be different, to leave possibilities open for future implementations.
Fine, but that was not the intent of my patch. The order is not different on any of the platforms as the (new) tests show.
If I do need to change the patch it will still need that broken() stuff I just added.
-
Paul Vriens -
Vincent Povirk