[Users] [Bug 2577] Focus rectangle on folder list and message list headings doesn't get properly ...

[Michael Shell]

On Wed, 22 Feb 2012 13:11:19 +0100
Andreas Rönnquist <mailinglists at gusnan.se> wrote:

> Using your "1-pixel" version of the fix works pretty good here too -
> with one small exception - If I step to the next message in the
> message list using the keyboard down key, that makes the next message
> highlight with double lines below - Which doesn't get erased properly
> when selecting another message.
> . 
> Is the same code used when stepping through the messages using the
> arrow keys as if using the mouse to mark a message?


  Andreas,

I went over it and over it and thought there was some kind of rounding
error in my focus rectangle code. Alas, this isn't the problem. In
fact, the phantom line isn't even from the focus rectangle - it's from
the background fill (dark blue) text rectangle. Note that the phantom
line remains dark blue even for red message titles (and red focus
lines). The 

cairo_stroke(cr);

code for the focus rectangle can be totally commented out and the
problem will still appear.

So, you have actually found a second bug!

When the mouse is used to select a message, all is well. When the
keyboard *up* arrow is used to select a message, all is well. However,
when the *down* arrow is used to select a message, the dark blue text
fill box will actually extend down one extra pixel from normal - and
that is what makes the phantom line. If you look carefully, you may
even see the dark blue "leaking out" 1 pixel below the focus rectangle
(this is easiest to see with a red message title, and thus red focus
rectangle) after the down arrow selects the message.
 
So, to reproduce this bug, use the down arrow key to advance downward to
select a message, then use the mouse to click on a message not adjacent
to the currently selected message (say, 2 messages up). The dark blue
phantom line will then appear. It will also happen with inboxes in the
folder window the same way.

I will try to look into this further and see if my eye catches any other
"off by 0.5 to 1.0 pixel" errors. Sigh.


  Cheers,

  Mike