https://bugs.documentfoundation.org/show_bug.cgi?id=70976
--- Comment #27 from libreofficeb...@toeai.com ---
I concur with Comment 20 that this seems to be related to fuzzy or naive
arithmetic. I did quite a few tests and observed the results. I won't bore
you with all the details, but in general I found:
1) I did not experience any desktop showing. The defects are these:
a) There are black lines of various widths on the outside always affecting
bottom and right but sometimes also left side.
b) Inside the black lines, there are sometimes white lines of various
widths on the bottom and right.
c) Inside the white lines if present, or inside the black lines if there
are no white lines, there are variously colored lines always 1px thick
affecting the bottom and right sides, or sometimes just the bottom or just the
right. When they are both present, they are not the same color as each other,
and the lines meet at a single pixel which is a different color from either
line. The colors of these lines are related to the slide's background color
and to the presentation size (see below), but I haven't found an exact formula.
Like the white lines, these lines are entirely inside any black lines, so if
the screen has a black line on the left, this colored horizontal line will not
extend into it.
2) The area of the screen affected (i.e. the widths and presence of the lines
described above) is dependent on the size of the presentation (i.e.
width/height in inches of the slide). The only thing that changes by changing
the background color is the color of the variably colored 1px lines.
3) The defect is magnified by decreasing the presentation size, minimized by
increasing the presentation size.
I have not yet found a size that makes the defect go away entirely. If your
background color is black, it may be a workaround to set a ridiculously large
presentation size, because even though the black lines will still be present,
they will be small and the white and variously colored lines don't seem to show
up then.
Here are the details from just four tests, going by orders of magnitude. The
background color in these tests was set to Indigo (0x55308d). Output screen
was 1366 x 768 pixels. Colors of variable lines given in RGB hexadecimal.
User defined size: 160"x90"
results:
1px black line on bottom.
2px black line on right.
User defined size: 16"x9"
results:
1px black line on bottom.
2px black line on right.
User defined size: 1.6"x0.9"
results:
1px black line on bottom.
2px black line on right.
5px white line horizontally and 3px white line vertically just inside the black
lines.
nexus pixel (0xf3f0f7) at inside corner of white lines joining 1px vertical
line (0xe1daeb) and 1px horizontal line (0xbbacd1).
User defined size: 0.16"x0.09"
results:
2px black line on left.
1px black line on bottom.
4px black line on right.
32px white line horizontally and 58px white line vertically just inside black
lines.
nexus pixel (0xf7f5fa) at inside corner of white lines joining 1px vertical
line (0xebe7f2) and 1px horizontal line (0xbbacd1).
Screenshots or other tests available on request.
Observed on:
OS: Kubuntu
screen size: 1366x768
version info:
-----
Version: 7.4.2.3 / LibreOffice Community
Build ID: 40(Build:3)
CPU threads: 4; OS: Linux 5.19; UI render: default; VCL: x11
Locale: en-US (en_US.UTF-8); UI: en-US
Ubuntu package version: 1:7.4.2~rc3-0ubuntu1
Calc: threaded
-----
This might seem to some like a "minor" bug of only "medium" importance, but
it's the sort bug that makes it hard to take LibreOffice Impress seriously (and
I really want to because I work on Linux mostly and because MS PowerPoint is a
very buggy expensive piece of junk). For a bug to go back nearly a decade now
and exist on all platforms? Not minor. How many people have been turned off
from using Impress just by observing that it can't even do the simplest of
things right (show me a blank screen), let alone reliably do anything
complicated or fancy?
And at ultra low presentation sizes (see above) we're not talking about a pixel
or two. 2+1+58+4 = 65 pixels! This isn't merely an "off by one" error. It
appears to involve someone making some super naive assumptions/calculations
about foundational geometry.
I'm not here to complain about free software. I'm trying to help see it
improved, if possible, so please let me know if I can provide more helpful
details.
--
You are receiving this mail because:
You are the assignee for the bug.
