I can confirm that kimageformats 6.16.0 gives the spurious error and 6.17.0 does not.
pm: reset flag term_force_init for each plot
It may be related to the below https://sourceforge.net/p/gnuplot/bugs/2397/ If I set the environment variable PANGOCAIRO_BACKEND=fc (fontconfig) it seems to be better.
Part of letters missing from ylabel/y2label
Thanks, setting this environment variable fixes this issue for me. 👍
It may be related to the below https://sourceforge.net/p/gnuplot/bugs/2397/ If I set the environment variablePANGOCAIRO_BACKEND=fc (fontconfig) it seems to be better.
Part of letters missing from ylabel/y2label
Part of letters missing from ylabel/y2label
The change should be in kimageformats-6.18.0, if I am reading the repository notes correctly.
gnuplot-6.0.x - qt terminal error "IFFChunk::innerFromDevice: unkwnown chunk"
After a system upgrade I'm getting this message also. II think it comes from one of the KDE Qt6 plugin libraries. On my system this symbol is found in kimg_iff.so, which is provided by kimageformats-6.16.0. Upstream fix may be this commit. I am not sure this actually fixes it; just sends the message via a different path. https://invent.kde.org/frameworks/kimageformats/-/commit/083680eb7729f02eb5d086782f45c042309d0911 chunks.cpp - send info and warning messages to the iff logging category For example:...
After a system upgrade I'm getting this message also. II think it comes from one of the Qt libraries. On my system this symbol is found in kimg_iff.so, which is provided by kimageformats-6.16.0. I can find a couple of references online to this as a Qt bug, including one that says it has been fixed upstream. Unfortunately no specific Qt versions were mentioned. If you figure which version it broke with and/or which version fixes it, please let me know so I can include this infor in the RELEASE_NO...
Thank you for the confirmation and thorough verification. Could you please specify which gnuplot version and which terminal types are in use when you see this bug? This may no longer be necessary, but I'll provide the information just in case: Terminal: x11 Version: master (commit: 3114766baee) Depending on the coordinate precision, the delta between the two curves may not go to zero. The test needs to allow for that. In my environment as well, it worked properly after considering the precision tolerance...
Fix incorrect above/below determination in polar filledcurves
Fix above/below treatment of filledcurves in polar mode
I am reclassifying this as a bug report so that when a fix is made it can refer to the issue by number. I now think that the apparent terminal-dependence is actually due to a difference in the precision of coordinates used by different terminals. There is a test later in the code that looks for points where two curves cross. Depending on the coordinate precision, the delta between the two curves may not go to zero. The test needs to allow for that. With that fix in addition to your code change I...
Ticket moved from /p/gnuplot/patches/830/ Can't be converted: _milestone: Version 6 _priority: 5
I am reclassifying this as a bug report so that when a fix is made it can refer to the issue by number. I now think that the apparent terminal-dependence is actually due to a difference in the difference in precision for the coordinates used by different terminals. There is a test later in the code that looks for points where two curves cross. Depending on the coordinate precision, the delta between the two curves may not go to zero. The test needs to allow for that. With that fix in addition to...
Could you please specify which gnuplot version and which terminal types are in use when you see this bug? Your patch looks reasonable to the eye, but there are some puzzling points. In the current development tip (6.1) I can sort of reproduce it in the qt terminal and in postscript output, but not for wxt or x11 or pdf output. The gd terminals produce something odd that doesn't match either your before or after screen shots. Since the problem (at least in the development branch) seems to be terminal-specific,...
Could you please specify which gnuplot version and which terminal types are in use when you see this bug? Your patch looks reasonable to the eye, but there are some puzzling points. In the current development tip (6.1) I can sort of reproduce it in the qt terminal and in postscript output, but not for wxt or x11 or pdf output. The gd terminals produce something odd that doesn't match either your before or after screen shots. Since the problem (at least in the development branch) seems to be terminal-specific,...
Fix incorrect above/below determination in polar filledcurves
Thanks for committing an improved implementation — much appreciated!
update INSTALL NEWS
polygon data cannot be contoured
avoid reference to uninitialized structure content
webp: support animation on a transparent background
kitty: support animation on a transparent background
kitty terminals: default to full terminal width , 3/4 height
"reset session" must terminate multiplot mode
backport fixes for pm3d fillcolors
"splot with circles" fillcolor lost if hidden3d is active
Add code in the sharpen filter to ensure the function is sampled at x=0
Add code in the sharpen filter to handle step functions
Add pre-defined variable Inf (INFINITY)
gprintf %H format in utf8 context now uses dot operator between mantissa and exponent
docs: cleanup (gprintf, geographic)
Remove EXPERIMENTAL warnings
Support use of either ZEXINT or CEXINT (Amos library expint routines)
svg: font size changes within a text fragment could be lost
resolve ambiguous syntax in "set dashtype"
cast colorspec.lt to (unsigned int) when it may hold 32-bit ARGB
polygon data cannot be contoured
Handle split contour lines as one continuous curve
Fixed by commit 83cd1370 Instead of building up a contour line using a fixed-length array of 100 points, dynamically allocate additional points as needed.
do not break contour lines into 100-segment fragments
avoid reference to uninitialized structure content
I think you may be overlooking a simpler explanation, and simpler fix, for the discontinuities seen in this example. The 100 point per contour segment is a defined constant. In your example case bumping this up to 1000 points removes the discontinuities. One possible fix is to make this a limit that the user can set. diff --git a/src/contour.c b/src/contour.c index 8a0f6a054..a87a758b3 100644 --- a/src/contour.c +++ b/src/contour.c @@ -79,7 +79,7 @@ typedef enum en_edge_position { #define FALSE 0...
webp: support animation on a transparent background
kitty: support animation on a transparent background
gd: further deprectate "optimize" option for gif animation
expand_unicode_escapes(): use memmove rather than memcpy because strings overlap
configure: default to --enable-function-blocks --disable-tektronix
Frozen plot window inside noVNC desktop
Dear @sfeam. You were completely correct that the issue with the window was due to the misproper window manager. I have solved the issue by making sure that the gnuplot module gets loaded inside the noVNC container (and not before launching it). Apparently, this makes a huge difference. Thanks again for your patience. This ticket can be closed if you wish. Regards Ehsan
Handle split contour lines as one continuous curve
Incorrect ARGB interpretation of color column in splot's boxes and polygons styles
NEWS needs to be updated (missing 6.x release info)
Do not disable mousing after a ^C
make sure multiplot mode is off after "reset session"
kitty terminals: default to full terminal width , 3/4 height
NEWS needs to be updated (missing 6.x release info)
The NEWS file is only updated for actual releases. You can find it in branch-6-0-stable. There is no need for in the development branch; it's just there as a place-holder for when a new stable series is started.
NEWS needs to be updated (missing 6.x release info)
I am pretty sure your problem is not with gnuplot per se, but with the lack of a window manager. I am not familiar with your container system at all, but I am guessing from the startup command you showed that it is starting up only a terminal (xfce-terminal) and not the actual xfce window manager. So you get a terminal window from which you can launch new commands that may open additional windows, but there is no way of managing those new windows because the other relevant xfce components have not...
Dear Ethan Thanks for your extended response. As a matter of fact, the pointer on switching to the wxt terminal could at least solve part of the problem. I built a custom gnuplot v6.0.2 coupled with wxWidgets module, and the resulting plot window can be only moved, but not minimized or closed. So, this is already some partial progress. Any more remarks on being able to control/customize the plot window under wxt? By the way, the 'q' hotkey works fine to quit the plot window. The other fallbacks (i.e....
kitty terminals: In "anchor" mode, clear terminal before plotting
You are describing a multi-layered environment that I have no experience with, so I can only offer some general suggestions or pointers. As you probably know, the lack of a frame for the plot window is due to the window manager (or lack of one). You mention xfce is available, but is it actually running? If so, why is it not providing a frame for the window? Moving/resizing/closing/raise/lower is also controlled by the window manager. However, the gnuplot plot windows for both x11 and qt should close...
label problem with pdfcairo
Associating this with #2048
Setting dashtype in loop only works for string values
fillcolor setting ignored in "splot with circles" when using "set hidden3d"
"with marks" always accept explicit linecolor in plot command
docs: typos
I found some points in docs/gnuplot.doc of git version at 2025-07-22, which may be typo.
japanese-20250731.tar.bz2: update gnuplot-ja.doc and term-ja.diff for git version at 2025-07-22.
fix color assignment of pm3d "fillcolor variable"
Frozen plot window inside noVNC desktop
Compilation with Qt6 is now supported with the MSYS2/MINGW64 and CLANG64 environments. See also [bugs:#2821].
fix color assignment of pm3d interpolate > 1 and "fc linestyle N"
fix color assignment of pm3d interpolate > 1 and "fc rgb variable"
Ticket moved from /p/gnuplot/bugs/2823/ Can't be converted: _milestone: _priority:
Security Alert Regarding Gnuplot Binary
Windows build: Recent change breaks old build chm file.
check for "fc background" in pm3d surface plots
comment out HHC for new help complier solves the issue
This is posted by me.
Windows build: Recent change breaks old build chm file.
Keyboard & mouse commands not working with Qt6 on Windows
.Turns out that the issue was the zero argument to QtLocalSocket::waitForReadyRead(). Changing this to a small number of milliseconds (>0) fixes the issue. Commit in master branch along with changes to config/mingw/Makefile to enable compilation with Qt6.
mingw: optional compilation with Qt6
mingw: alternative help compiler
mingw: optionally use the MSYS2/CLANG64 environment
qt: fix for Qt6 on Windows
The failing communication path is gnuplot receiving events from gnuplot_qt. There's never a call to qt_processTermEvent(), whereas on the sending side QtGnuplotEventHandler::postTermEvent is called and writes to the socket. Adding a flush() after write() does not help. Interestingly, receiving fontprops in qt_sendFont() works. But the code in qt_waitforinput() which is supposed to read events from the gnuplot console, or the wgnuplot text window, and the windows or wxt terminals, and the qt socket...
The failing communication path gnuplot receiving events from gnuplot_qt. There's never a call to qt_processTermEvent(), whereas QtGnuplotEventHandler::postTermEvent is called and writes to the socket. Adding a flush() after write() does not help. Interestingly, receiving fontprops in qt_sendFont() works. But the code in qt_waitforinput() which is supposed to read events from the gnuplot console, or the wgnuplot text window, and the windows or wxt terminals, and the qt socket fails to receive any...
Both gnuplot and gnuplot_qt send and receive to each other, so there are four communication paths. I am trying to understand or help diagnose which of them is failing. If plotting works, I take it that means commands are successfully sent from gnuplot to gnuplot_qt. When you say keyboard commands do not work, what does that mean exactly? Commands after the initial "plot" command are ignored? Keyboard is locked up? New commands work but hot-keys (bind commands) do not? Does the plot window (gnuplot_qt)...
Keyboard & mouse commands not working with Qt6 on Windows
Turns out that compilation with Qt6 is straight forward with some minor tweaks to the Makefile only. But indeed the communciation channel (using Windows specific code) no longer works. Plotting works fine, but keyboard/mouse commands are not sent/received. That was some really complicated code to get right in the first place. I make no promises as to when I will get around looking at this.
I agree that switching official Windows builds to Qt6 is eventually a good thing. It still runs fine with Qt5, though, and there's a bit of Windows-specific trickery to enable the wgnuplot GUI and windows, qt, and wxt terminals to work at the same time. That may or may not work with Qt6 atm. At the minimum, the Makefiles in config/msvc and confing/mingw would need an update. It is easiest to build gnuplot with MSYS2/MINGW64 on Windows as MSVC library dependencies can be a pain. (Also, I would recommend...
I've noticed that since v5.4.9 Qt6 has been supported, but the recommended usage for Windows users is to get the zip or 7z folder to use the binaries, which still uses Qt5 dll's. I tried the other suggested method for building on Windows (running nmake in the config/msvc directory), but kept running into roadblocks only to discover that it looks like Windows Qt6 support hasn't been included. This is potentially a bad thing for projects that use Windows and Linux, because if Linux user's build with...