We're currently only using GimpHighlightableButton in the layers
dialog, which defines its own set of highlight colors. We're going
to use highlightable buttons in the dashboard too, so let's move
the highlight colors to gimphighlightablebutton.h, and give them
standard names. We currently define
GIMP_HIGHLIGHTABLE_BUTTON_COLOR_AFFIRMATIVE (green), and
GIMP_HIGHLIGHTABLE_BUTTON_COLOR_NEGATIVE (red).
(This commit was accidentally dropped from the gimp-2-10 branch; it
should have gone before 40ac4f7bc0f43aee24dc7ae1cf674d1a59612f55.)
performance-log-expand.py decodes a delta-encoded performance log
by expanding the deltas, producing a log where each sample (and
other relevant elements) contain complete information. Note that
the structure of expanded logs is identical to that of delta-
encoded logs, the expanded log simply has no deltas.
performance-log-resolve.py resolves symbol information in
backtraces. The logs produced by GIMP only specify the program
counter at each stack frame, providing an address-map to map
program-counter addresses to actual symbols separately. This tool
looks up each program-counter address in the address map,
incorporating the relevant symbol information directly into the
backtrace.
Both tools read their input from STDIN, and write their output to
STDOUT, and can be chained in a pipeline (with
gimp-performance-log-expand.py appearing first).
Note that these tools require Python 3.
(cherry picked from commit d7c74a615b)
Add an option to record a performance log through the dashboard.
The log contains a series of samples of the dashboard variables, as
well as the full program backtrace, when available. As such, it
essentially acts as a built-in profiler, which allows us to
correlate program execution with the information available through
the dashboard. It is meant to be used for creating logs to
accompany perofrmance-related bug reports, as well as for profiling
GIMP during development.
The sample frequency defaults to 10 samples per second, but can be
overridden using the GIMP_PERFORMANCE_LOG_SAMPLE_FREQUENCY
environment variable. Backtraces are included by default when
available, but can be suppressed using the
GIMP_PERFORMANCE_LOG_NO_BACKTRACE environment variable.
Logs are created through the new "record" button at the bottom of
the dashboard dialog. When pressed, a file dialog is opened to
select the log file, and, once confirmed, data is being recorded to
the selected file. Recording is stopped by pressing the "record"
button again (we use a highlight to indicate that recording is
active.)
While recording, the "reset" button is replaced with an "add marker"
button, which can be used to add event markers to the log. These
can be used to mark events of interest, such as "started painting"
and "stopped painting", which then appear in the log as part of the
sample stream. Markers are numbered sequentually, and the number
of the next (to-be-added) marker appears on the button. Shift-
clicking the button adds an empty (description-less) marker, which
is only identified by its number; this can be used when markers
need to be added quickly.
The log is an XML file, containing some extra information (such as
the output of "$ gimp -v", and symbol information) in addition to
the samples. The data in the file is delta-encoded to reduce the
file size, meaning that samples (as well as some other elements)
only specify the changes since the previous sample. This adds a
necessary decoding step before data can be processed; the next
commit adds a tool that does that.
There are currently no tools to actually analyze the data -- that's
still TBD -- but at least we can start gathering it.
GimpBacktrace provides an interface for creating and traversing
multi-threaded backtraces, as well as querying symbol information.
While we already have some backtrace functionality, it relies on
external tools for the most part, and as such is rather expensive,
and is only meant for producing opaque backtraces. GimpBacktrace,
on the other hand, is meant to be relatively cheap (we're going to
use it for profiling,) and allow inspection of the backtrace data.
In the future, it might make sense to replace some, or all, of the
other backtrace functions with GimpBacktrace.
GimpBacktrace currently only supports Linux. By default, it uses
dladdr() to query symbol information, which is somewhat limited (in
particular, it doesn't work for static functions.) When libunwind
is installed, GimpBacktrace uses it to get more complete symbol
information. libunwind is currently an optional dependency, but it
might make sense to promote it to a mandatory, or opt-out,
dependency, as it's lightweight and widely available.
On other platforms, the GimpBacktrace interface can still be used,
but it always returns NULL backtraces.
(cherry picked from commit 80bf686c94)
Move the call to gimp_filter_tool_disable_color_picking() from the
filter-tool's dialog "unmap" handler to gimp_filter_tool_halt().
Since commit ec80a88513, we
explicitly destroy the GUI when halting the filter tool, which
happens before the dialog's unmap handler is called, which could
potentially result in a dangling pointer to the active color-picker
widget in gimp_filter_tool_disable_color_picking().
(cherry picked from commit 072d6b0d12)
... 100% position anymore
In GimpGuideTool, use a closed [0, max_position] range as the
allowable range for new/repositioned guides (where max_position is
either the image's width or height), so that guides can be placed
at the right/bottom edge of the image.
(cherry picked from commit 547190faa8)
In gimp_image_merge_layers(), explicitly fetch the graph of the top
layer's parent layer (if exists), to make sure that the top layer's
graph has a parent node. We already fetch the image graph, which
takes care of top-level layers, however, if the top layer is a
child of an invisible layer group, as is the case in the wavelet-
decompose plug-in, this is not generally enough to guarantee that
the group's graph is constructed.
(cherry picked from commit e563845174)
In gimp_filter_tool_halt(), explicitly clear the GUI container
before clearing filter_tool->config, since the tool might be halted
during the GUI dialog's delete event, in which case the GUI will
only be implicitly destroyed *after* the function returns. The
destruction of the GUI might fire signals whose handlers rely on
filter_tool->config, so we need to make sure it happens while it's
still alive.
In particular, this fixes a CRITICAL in the threshold tool, which
occurs due to the histogram view's "range-changhed" signal being
fired during its destruction, and its handler accessing
filter_tool->config.
(cherry picked from commit ec80a88513)
gimp_device_info_set_device(): don't just bail out if a device with
the same name is added again, instead, simply continue and overwrite
the info's old device with the new one.
NOTE that this only happens if something is wrong on the USB or udev
or libinput or whatever side and the same device is present multiple
times. The only "safe" thing is to assume that devices listed earlier
are dead and dangling entities and that the last registered device is
the one actually delivering events.
(cherry picked from commit 717c183a3e)
In GimpProjection, use an adaptive chunk size when rendering the
projection asynchronously, rather than using a fixed chunk size.
The chunk size is determined according to the number of pixels
processed during the last frame, and the time it took to process
them, aiming for some target frame-rate (currently, 15 FPS). In
other words, the chunks become bigger when processing is fast, and
smaller when processing is slow. We're currently aiming for
generally-square chunks, whose sides are powers of 2, within a
predefined range.
Note that the chunk size represents a trade off between throughput
and responsiveness: bigger chunks result in better throughput,
since each individual chunk incurs an overhead, in particular when
rendering area filters or multithreaded ops, while smaller chunks
result in better responsiveness, since the time each chunk
individual takes to render is smaller, allowing us to more
accurately meet the target frame rate. With this commit, we aim to
find a good compromise dynamically, rather than statically.
The use of adaptive chunk sizes can be disabled by defining the
environment variable GIMP_NO_ADAPTIVE_CHUNK_SIZE, in which case we
use a fixed chunk size, as before.
(cherry picked from commit a1706bbd29)
Add a boolean "chunk" parameter to
gimp_projection_chunk_render_iteration(), which determines whether
the work area should be sub-divided into chunks prior to rendering
(previously, the work area would always be sub-divided.) Only
pass TRUE when rendering the projection asynchronously, in the
render callback, and pass FALSE when rendering the projection
synchronously, in gimp_projection_finish_draw(), which is called
when flushing the projection through the GimpPickable interface.
Rendering the projection using as big chunks as possible improves
performance, while worsening responsiveness. Since responsiveness
doesn't matter when rendering synchronously, there's no reason to
render in chunks.
(cherry picked from commit 105ffc787d)
In gimp_transform_tool_transform(), use "active_item", instead of
"tool->drawable", when cutting/pasting the selected portion of a
layer for transformation. The latter is a remnant of the old
transform-tool code, and is not guaranteed to be correspond to the
correct drawable, or even to a valid drawable (i.e., it can
potentially produce wrong results, or segfault.)
(cherry picked from commit 9420805525)
In the command handler of GimpTileBackendPlugin, forward unhandled
commands to gegl_tile_backend_command(), instead of asserting that
they're within range (which has already been disabled by commit
bc3b076caf). See GEGL commit
30047e65723ebb44fcde9c6b5f60ceecb43b0895.
(cherry picked from commit 668fee966a)
... which reports the amount of data queued for writing to the
swap (see GEGL commit 64021786ee067cf66c038622719acc590e6341db.)
When the swap queue is full, a yellow color underlay is shown in
the history graph.
(cherry picked from commit cd54457d46)
The "compression" field reports the ratio between the total size of
the data in the swap, and the total size the data would have had if
all tiles in the swap occupied a unique data block.
See GEGL commit 185f4450f2a51690b39112973c61f894c1ec3e41.
(cherry picked from commit b6e552a74b)
In gimp_tile_backend_plugin_command(), disable the range check for
the input tile command. This check prevents us from adding new
tile commands to GEGL without breaking the ABI; yet, the next GEGL
release will add a new command. We're going to have to decide what
to do about this, but for now, let's just disable the check, so
that at least GIMP 2.10.6 is compatible with newer versions of
GEGL, no matter how we end up handling this.
Do not set the interpreter to `python2` but to whatever was found by the
AM_PATH_PYTHON2() m4 macro.
It looks like the Python2 binary we ship in our DMG may be call "python"
only (without the '2'). Let's just make our code more resilient to
various builds.
I am not sure yet this is the only/actual problem for this issue on
macOS, but this is at least in the right way.
If I override the `program` variable, and it is not found in PATH
environment, then it is NULL and the error message is unhelpful. Make
the return value of g_find_program_in_path() into a separate variable
instead, and only override `program` in the end, when we know it is
non-NULL.
The python path with directory are generated files and should not be in
the dist. The previous commit was fixing an in-tree `make check` but
this one fixes the `make distcheck` as the dist was packaging the wrong
python files.
This issue was invisible when making VPATH builds, but appeared only in
source-tree builds as gettext tools are apparently mixing source and
build files. So we need to add these in the POTFILES.skip.
