aboutsummaryrefslogtreecommitdiff
path: root/README.md
diff options
context:
space:
mode:
authorLukasz Kasprzak <lukasz@arcofasiagroup.com>2026-08-21 09:49:33 +0200
committerLukasz Kasprzak <lukasz@arcofasiagroup.com>2026-08-21 09:49:33 +0200
commit7459e218fa8b5b86c059762519447cb8863d3e75 (patch)
tree413f960c8543a0d8016c2ce45132648c61e25b6e /README.md
parentd58ac7945a97b4029772022550775ab834c81b97 (diff)
downloadcolitur-x-7459e218fa8b5b86c059762519447cb8863d3e75.tar.gz
colitur-x-7459e218fa8b5b86c059762519447cb8863d3e75.zip
debug: MUX_DUMP writes one frame from the pixmap, then exits
Verifying what the program actually draws by screenshotting its window is unreliable: X does not retain a window's contents while it is obscured or on an unfocused tag, so import and xwd can both return solid black while the rendering is correct. Reading back the off-screen pixmap always works. Used to confirm the XRectangle clip-truncation fix: rows now run unbroken from 2027-01-01 to 2027-01-29 with every column populated, where before the twelfth row lost everything but its name.
Diffstat (limited to 'README.md')
-rw-r--r--README.md11
1 files changed, 11 insertions, 0 deletions
diff --git a/README.md b/README.md
index 920c056..b0872cf 100644
--- a/README.md
+++ b/README.md
@@ -76,6 +76,17 @@ and everything after it silently disappears, which looks like a layout or
buffer-overflow bug and is neither. `set_clip()` clamps to the buffer bounds
before converting.
+### Verifying a frame without a screenshot
+
+ MUX_DUMP=/tmp/frame.ppm ./colitur-mu # writes one frame, then exits
+ convert /tmp/frame.ppm /tmp/frame.png
+
+This reads back the off-screen PIXMAP, not the window. A window's contents
+are not retained while it is obscured or sitting on an unfocused tag, so
+`import`/`xwd` against one can return solid black while the program is
+drawing perfectly -- which cost real time to work out once. A pixmap is
+always there.
+
### Which to use
`colitur-x` renders; `colitur-mu` renders and lets you change the invocation.