The old algorithm treated the "Words" checkbox as meaning match the
entire text, not match words, which is unlikely to be what anyone is
expecting.
Also fixes replace to replace all instances in a single text item
rather than just the first.
Reported on the forum.
Text has a fairly big bounding box to account for descenders, overbars,
etc., but it makes it feel too sloppy for hittesting. This change
allows selection disambiguation to look at the actual strokes of the
text when deciding what's "closest".
Fixes https://gitlab.com/kicad/code/kicad/issues/9506
ADDED: Option to assign unique reference designators to footprints
created with the array tool [pcbnew]
CHANGED: Array creator dialog no longer displays pad numbering properties
when in the board editor. [pcbnew]
Fixes https://gitlab.com/kicad/code/kicad/-/issues/2354
We no longer have any knowledge of the original start/end of the arc,
since SetArcAngleAndEnd swaps the start and end to ensure the arc is
always clockwise at the end.
Adds a method EDA_SHAPE::EndsSwapped() to notify whether the start/end
point of the shape were swapped.
Make auto save recover all auto saved sheet files not just the root sheet.
This does not fix existing auto save issues because recovery would require
an iterative reload for each automatically saved sheet file and would only
happen once.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/9288
Rather than detaching and re-attaching the layer panels with the
attendant overhead, we keep each panel intact and only update its
contents. Since the bindings are the same, this saves substantial time
on pcb loading as well as when closing the board setup dialog
Unused variables in function calls can be commented out. Unused
return variables get a new routine `ignore_unused()` that silences the
warnings with zero overhead.
Also cleans up the dialog a bit.
Also changes the workings of NormalizePath to return the input if
it couldn't be shortened which saves open coding that behaviour in
all its callers.
Fixes https://gitlab.com/kicad/code/kicad/issues/9036
The default is usually wrong and presenting the basic footprint
information when creating an empty footprint allows the user to
establish their preference before mistakes occur.
Fixes https://gitlab.com/kicad/code/kicad/issues/9215
Also removes the message bar display of assigned netclass for
buses and bus-to-bus entries.
Also fixes a bug where assigning a netclass via the canvas only
looked at the first level of bus members (and not any nested
members).
Also fixes a bug where the bus name validator tried to validate
a vector bus first -- which doesn't work as a vector bus may be
nested in a group bus.
Also fixes a bug where we were failing to check for illegal
chars in bus definitions which otherwise passed the bus parsers.
See additional comments in the bug report.
Fixes https://gitlab.com/kicad/code/kicad/issues/9160
Also removes the return value from the BITMAP_STORE's themeChanged
method as it will provide the wrong answer for the second frame it's
called on.
Also hooked up ACTION_TOOLBAR, the app launcher, and the project tree
to the wx event for system color changes so they change on the fly.
Fixes https://gitlab.com/kicad/code/kicad/issues/6026
Mac already uses diverse font sizes (for instance, smaller fonts for
radio button groups), and the anti-aliasing is a lot better. The
other platforms need a more limited range of sizes.
Fixes https://gitlab.com/kicad/code/kicad/issues/8608
This is a test split per requests. While "Latin American Spanish" is
not really homogeneous, there are distinct differences to European
spanish, so we are splitting in an effort to allow both translations.
If one is less viable, this may re-merge in the future.
Moving points should not be overrided by disambiguation menu, so we
track where the disambiguation event started and only trigger the menu
if the mouse hasn't moved a full unit in x/y
Fixes https://gitlab.com/kicad/code/kicad/issues/9132
Previously, we were checking if the selection tool was the top of the
stack but this ignored many other "secondary" tools that might be
running such as the point editor. These still allow the selection tool
to handle events such as clicks. This change allows the selection tool
to handle clicks when it is the primary tool on the stack rather than
the top
Fixes https://gitlab.com/kicad/code/kicad/issues/9110
Previously, the count of segments used a magic number optimized for Pcbnew.
This is not good, and does not work well on Gerbview.
The count uses now a max error acceptable is approximation (5 microns in Gerbview and Pcbnew).
Fixes#9101https://gitlab.com/kicad/code/kicad/issues/9101
This standardizes selection tools in SELECTION_TOOL class to be able to
handle a long-click as asking for a non-heuristic disambiguation menu.
Fixes https://gitlab.com/kicad/code/kicad/issues/8490
This assigns the "Tab" key to a general pcbnew 45° limitation. Limit
can be enabled/disabled using context menu as well.
Removed the user preference for 45° limitation on graphic lines as that
is overlapping with the general limitation, which is not an editing
preference so much as a temporary adjustment to tool action.
Related to https://gitlab.com/kicad/code/kicad/-/issues/8490
This function attempted a poor-man's natural compare but it assumed
specific structure of the string. This broke for strings with
numberings that looked like decimals.
Instead, we use our natural string comparison function and remove the
references to this older function and its similar elements.
Fixes https://gitlab.com/kicad/code/kicad/issues/9067
curl.h has constraints on Windows (because it includes winsocks2.h) and must be
included before any wxxxx.h.
Because only very few files need curl.h, kicad_curl.h is included only in these files
Because the same parser is used to read all footprint files of a library,
the parser must be reinitialized (internal variable cleared) before reading
a new file, otherwise the previous parser state is applied to the next file.
Fixes#7627https://gitlab.com/kicad/code/kicad/issues/7627
For historical reasons they are stored in a string using spaces as separators.
So each fp filter is now escaped to remove spaces (replaced by {space})
Fixes#9009https://gitlab.com/kicad/code/kicad/issues/9009
NETCLASSes are stored as shared pointers. We lose the reference as soon
as the pointer count decrements to 0, which can happen immediately when
the shared pointer is returned.
The lemon parser doesn't lend itself to short-circuiting booleans,
but if we defer processing until we fetch the second value when
processing the TR_OP_BOOLEAN opcode then we get to piggy-back on
C++'s boolean short-circuiting.
1) Don't wait for UpdateUserInterface to build the (global) list of
layer names.
2) Report clearance resolution between a silk layer and a mask layer.
3) When writing DRC reports, include info about the violating rule.
4) Report "no relevant layers" if we fail to find anything to write
a clearance resolution report about.
Fixes https://gitlab.com/kicad/code/kicad/issues/8963
Also reduces line-wrapping a bit.
I don't think this fixes the linked bug, but I can't reproduce it
and this was found while reviewing the code.
Fixes https://gitlab.com/kicad/code/kicad/issues/8810
They don't define a KiCad string class, so the header file name was
somewhat misleading. But the fact that they didn't match definitely
made coding more difficult.
This adds an icon to the left of the row with an error symbol if the
file can't be found or opened, and shows a tooltip over the icon with
and error string.
Fixes https://gitlab.com/kicad/code/kicad/issues/3815
Note that the dialogs which have mutliple options were not moved over
as it seems perahps not the right choice to have them save state between
sessions.
Fixes https://gitlab.com/kicad/code/kicad/issues/8762
The first time a DIALOG_SHIM was created, with a given size, this size
was not taken in account. This this is now forced.
(when recreated during a session, this is the last dialog size that is used)
Calling wxDialog::EndModal() directly from inside the dialog is a
potential bug if the dialog is shown in the modeless or window modal
(quasi-modal). Use the internal button events where possible and
check for the appropriate mode before calling the correct end dialog
function.
Found via `codespell -q 3 -S *.po,./thirdparty -L aactual,acount,aline,alocation,alog,anormal,anumber,aother,apoints,aparent,aray,dout,einstance,modul,ot,overide,serie,te,,tesselate,tesselator,tht`
m_shapes now has two possible indices. The first one is populated if
the point is associated with an arc and the second index is populated
if the point is shared between two arcs.
Also fixes a bug where an empty window would get left when loading
a file from the KiCad manager window errored out.
Also fixes a bug where we were getting the wrong exception because
we were rethrowing the type-casted one (rather than just calling
throw).
Fixes https://gitlab.com/kicad/code/kicad/issues/8604
It's tempting to say that we don't need to exclude filename chars
from symbols, but we might regret that decision down the road. Better
to just escape them.
Fixes https://gitlab.com/kicad/code/kicad/issues/8694
This allows us to avoid the monospaced font (along with the tab-stop
handling), and process a <return> as an OK rather than just <shift> +
<return>.
Fixes https://gitlab.com/kicad/code/kicad/issues/8425
(The current one keeps pulling the Kicad Manager window to the front
on OSX.)
I also removed the UpdateUI stuff which was probably causing:
Fixes https://gitlab.com/kicad/code/kicad/issues/5049
The max-char-width impl was buggy and never applied to PCBNew anyway.
Plus the stroke font has a max-char-width of 2.8 time the space width,
so it made for really wide tab spacing.
Also fixes a bug where the Scintilla editors weren't getting a
monospace font on Mac. (This bug may also exist on the other
platforms, but each would need its own fix.)
Also moves more of the Scintalla customizations to SCINTILLA_TRICKS
where they can be shared.
Fixes https://gitlab.com/kicad/code/kicad/issues/8666
We lock the project file, board, and schematic now, so the second copy
can only open them read-only. While issues might remain, it's unclear
the warning ever prevented anyone from shooting themselves in the foot
anyway.
Fixes https://gitlab.com/kicad/code/kicad/issues/7465
- Remove outdated y-offset that prevented restoring full-height
(client area returned by wx is correct these days)
- Clamp frame geometry to fit new display when display is changed
- Handle when an automatic window reposition causes a DPI change
Fixes https://gitlab.com/kicad/code/kicad/-/issues/8341
Reduce each (Accel & Fallback) to 3 options: Off, Fast, Good. Fast AA
in accelerated canvas uses SMAA with tweaks suggested by David Beinder.
Good AA is super-sampled x2.
Cairo is similarly reduced to Off, Fast, Good, which map to the Cairo
options themselves. Best is removed as it mostly affects text rendering
and not line drawing (as our text is)
The same logic block to resolve possible symlinks exists in 5 different
places now. This change moves this duplicated code to a static member
function of WX_FILENAME instead.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/8522
Conservative makes SMAA usable in Eeschema through weaker AA.
Aggressive reduces some artifacts compared to the High preset.
Restores the unused Ultra preset in the shader source to its original state.
Avoid fractional pixel offsets by forcing integer division of m_screenSize.
Shift rounded lookAtPoint onto OpenGL pixel center to ensure consistent rounding to int.
The pixel 0..1 @ 1x needs to fill the area 0..2 @ 2x so its center moves by 0.5 * (pixel size @ 1x).
This was leaking windows headers and partial wx headers to 1084 compilation units......
This also means math/util.h is leaking to 1084 compilation units which seems a bit high too.
This involved adding some extra infrastructure to be able
to handle the case where a track that is sitting in the
router's commit waiting to be added to the board actually
needs to be deleted instead.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/8419
CHANGED: Highlight net tool changed to highlight display toggle; moved to left toolbar
CHANGED: Multi-net highlight is now accessible when using the highlight net action
ADDED: Highlight Net and Clear Highlight actions to Net Tools context menu
ADDED: Default hotkey (Ctrl+`) for toggling net highlighting
The current and previous net highlight state are now stored and can be toggled
on or off with an action (no hotkey by default).
Highlight is now treated as a persistent state rather than a tool that is entered,
meaning there is no longer a special tool to enter in order to highlight nets.
Highlighting can be controlled while any other tool is active using the hotkeys
or context menu on a net object.
Tweak the icons for highlight net and layer display mode to better disambiguate them.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/8250
Differentiates better between the EDA_ITEM IsModified(), referring to
items themselves changing and the EDA_SCREEN IsContentModified(),
referring to whether we have made any unsaved changes.
Initialized COLOR4D takes an EDA_COLOR_T enum but then referenced it as
an index to the color table. This uses the enum as comparison against
the table-stored enum.
Adds standard orange to picker swatches
* Remove unused wxLogDebug calls.
* Add add application level character tracing to detect unhandled key
events.
* Catch unhandled exceptions in KiCad main event loop like single top
applications.
Note: PGM_BASE::InitPgm() is creating a SETTINGS_MANAGER object, which
in turn tries to display a "migrate settings" dialog when no settings
are found, but obviously the dialog can't be displayed inside docker.
Needed to add an optional parameter to InitPgm() in order to allow
running in headless mode.
Save sheet and symbol instance information to the clipboard on copy
Load sheet and symbol instance information from the clipboard on paste
and renumber page numbers after loading.
Correctly handle pasting in a multiple hierarchy by ensuring symbol and
sheet instances are updated for all instances of the destination sheet.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/8207
CHANGED: Environment variables defined outside of KiCad will no longer
be saved in the settings file. Paths can be configured via
the Configure Paths dialog and those changes will be saved in
the settings file if the new path is different from the default
and the path variable was not defined in the system environment.
wx/wx.h includes all wxWidgets .h files, and sometimes creates collision
names in #define between kicad and windows headers
Moreover, blindly including a lot of useless files is compil time consuming
REMOVED: Update schematic option from geographical annotation dialog due
to potential issues with incomplete and/or broken updates. Use "Update
Schematic from PCB" tool to update reference designation changes.
Forcing the footprint reference designator changes back to the schematic
without checking any other board changes in the schematic could leave the
schematic in a undefined state. The update schematic from board tool is
the correct method to sync any changes from the board.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/8042
On Linux, the documentation and help files are potentially installed to
a non-standard location (i.e., outside of /usr/share/doc/kicad/). This
can be the case when, e.g., multiple versions of KiCad are installed in
parallel. Making KICAD_DOCS available at run-time is the only viable
solution to allow the applications to find the help files in this case.
Fixes https://gitlab.com/kicad/code/kicad/issues/7874
Rely on existing code to initialise the list of paths in which the KiCad
documentation might be located, thereby making SearchHelpFileFullPath
platform-agnostic and easier to read.
On MSW, CaptureMouse() was sometimes called event if the GAL panel has
captured the mouse (that was not accepted by wxMSW), although HasCapture()
returns false.
(Perhaps some race condition between events)
The fix add a flag to avoid calling twice CaptureMouse() after only one
wxEVT_MOUSE_CAPTURE_LOST event was received.
Fixes#8239https://gitlab.com/kicad/code/kicad/issues/8239
Keeping them in base_units means that we can't introduce
any dependence on these functions to anything that needs to
compile without one of the unit defines (EESCHEMA, PCBNEW, etc)
Introduce a new advanced config variable `KICAD_LIBRARY_DATA` which can
be used to move templates, symbols, footprints, and 3dmodels out of
`KICAD_DATA`. If not defined, everything is kept as before.
To facilitate this, PATHS::GetStockEDALibraryPath() is added. This
allows to differentiate code paths looking for EDA library data vs. code
paths looking for plugins, demos, and the like.
Thanks to Aimylios for the hints and suggestions with regards to the
stock EDA library data path handling on Windows and MacOS.
VIEW::GetBoundary() returns the entire view area, not the visible area.
Surprisingly, we had no API for this, so I added one.
Also, changed the dragger behavior to toggle between optimizing just the
modified area and optimizing the visible area, i.e. we will now never
optimize off-screen portions of the dragged track.
This code was shared between the footprint frame and
pcb edit frame, so push it into the base edit frame.
Also remove the dummy wxUpdateUIEvents that weren't
actually being used.
Use the connecting straight tracks even if not exactly parallel - allow
an error margin configurable in ADVANCED_CFG (default 1 degree). Also
be less strict about end point matching and use the width of the track
as the criteria to determine suitability.
Finally, delete any short lengths of track at the end of the operation
and amend the arc end points to keep connectivity.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/7967
The native cursor must be set on the GAL canvas wx object,
not its frame (otherwise on GTK the cursor is only set
when we get a request for it from the WM and not immediately).
Fixes https://gitlab.com/kicad/code/kicad/issues/6421
This is under the supposition that we can't set the column widths
on some Mac instances because the host controls haven't yet been
created. This is primarily conjecture based on looking at things
that have the *possibility* of going wrong. Why this only happens
in some installs is beyond me.
Fixes https://gitlab.com/kicad/code/kicad/issues/5479
After commit 9535153f9e there were no more IDs inside the legacy ID
system. Therefore the entire system for dispatching and looking up
the legacy ideas and handling those events can now be removed and
the tool dispatcher simplified (it no longer needs to know about
the ACTIONS class).
This is hidden behind an advanced configuration setting and is primarily
useful for developers trying to troubleshoot the netlist payload sent
from the board editor to the schematic editor.
Fixes https://gitlab.com/kicad/code/kicad/issues/8051
Saying "Import/Export" on the labels when the submenu has "Import"
or "Export" in it is redundant.
Note that we can't just update the action text with the new name,
because that is used in the hotkey list and would become too confusing
without the "Import"/"Export" text.
Before, the layers were always ellipsized if the name was >30
characters, even if the widget was wide enough to display
the full name. Now they are ellipsized when needed (e.g. if
the widget is shrunk small enough).
Fixes https://gitlab.com/kicad/code/kicad/issues/1919
The text areas aren't always redrawn when the window is activated
or deactivated, so we need to force a refresh of the text to ensure
it is the correct font color.
This allows for color changes and scaling of the arrow
when the display needs it (e.g. HiDPI displays).
Also make the arrow and text respond to if the window is
active to mimic native controls.
Fixes https://gitlab.com/kicad/code/kicad/issues/6042
Action plugin authors can specify alternative icon path for dark theme.
If it's not specified then standard icon will be used for both light and
dark themes. If neither icon is specified then the default puzzle piece
icon is used.
Fixes#7984
It is not always useful to distinguish between the cases when KiCad is
run from the build or the installation directory. E.g., the libraries
are generally not available in the same path as the code. Make the
evaluation of KICAD_RUN_FROM_BUILD_DIR conditional to accomodate these
scenarios.
1) Accept <enter> as OK if the multiline textbox isn't shown.
2) Focus and select all in first control if multiline textbox isn't shown.
3) Make title be title caps ("Edit Sheet name Field" looked dorky).
Fixes https://gitlab.com/kicad/code/kicad/issues/7940
If we keep doing design validations at two different places, they can
only start to drift apart. Board Setup should validate for malformed
data ONLY; everything else is the business of DRC.
This also fixes a bug where implementation validations would not allow
OK in the dialog, but wouldn't put up a message to say why. We now use
the InfoBar impl from Jon's commit for that.
Bitmaps are now identified by an enum class instead of by pointers.
Bitmap loading and caching is now handled by a class in common, and
we no longer compile most bitmaps into the binary, so there is no
longer a bitmaps static library.
Instead, bitmaps are archived to a .tar.gz file which is installed
in ${KICAD_DATA}/resources/images.tar.gz
The source PNGs are checked in to Git as the original CPP files were,
so that people can build without the required dependencies to convert
SVGs to PNGs.
Initial support is also added for dark theme icons, although this
is not yet exposed in the GUI.
Stubs are present for multi-resolution image resources, but this is
not fully-baked yet and could use some refinement.
Updates the Serbian language code from sr_RS to sr. This appears to
work as intended under Debian whereas the full code, despite existing in
the i18n directory was not recognized by wxWidgets or Weblate.
When updating the libtree, we modify the tree multiple times to avoid
crashes on Linux. These generated events that buffered while the widget
was frozen and resulted in redrawing the same footprint 4-10x.
This reduces the buffering by discarding events that are generated
during the freeze and only redrawing when there is a new footprint to
draw
The move tool can stack on others, so when we re-enter the previous tool
this caused another immediate action. We flag re-entry in the tool
stack to check for this and avoid unexpected tool starts
Vertex memory allocation uses malloc and realloc without checking the
result. Throw exceptions that can be caught in the DoRepaint() method
in EDA_DRAW_PANEL_GAL when malloc and realloc return a null pointer.
Add a reset button that will only show up if the user
edits the JSON configuration for grids. This is a temporary
measure until we have a real grid editor GUI in V7.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/7595
- remove ${COMMENT0}, that was broken (returned the same as ${COMPANY}.
- So ${COMMENT1} to ${COMMENT9} are only recognized for Comment1 to Comment9
in page settings dlg, and are equivalent to our old %C0 to %C8 format string
- fix a minor coding style issue.
Fixes#7642https://gitlab.com/kicad/code/kicad/issues/7642
Each symbol unit in eeschema has a unique identifier. But we don't have
a unique identifier for the entire symbol. So changing which symbol
instance was unit A (our default base for matching), changed the UUID
that we were using to match the footprints.
This commit adds all UUIDs to the netlist, allowing us to match symbol
to footprint without worrying about which unit is referenced.
This still does not handle changing different units on different sheets.
Fixes https://gitlab.com/kicad/code/kicad/issues/7604
The proposed overload did not compile on gcc and using adjustments to
get to compile by moving definition into cpp results in false positives.
This is removed until we can come up with a better solution and test
- Make DIALOG_SHIM::SetPosition working)
- rename PCB_BASE_FRAME::SelectLayer() to PCB_BASE_FRAME::SelectOneLayer()
- PCB_BASE_FRAME::SelectOneLayer(): make dismiss dialog by escape key working.
Fixes#7578https://gitlab.com/kicad/code/kicad/issues/7578
For some kinds of issues, we want a way to point them out but
we don't need to keep the user from being able to close the
dialog. A separate Validate button lets the user check for
issues if desired, but these issues are not "fatal" and can
be ignored if the user wishes.
The automatic wire tool was not working in the gridless mode because the
grid snapping on/off logic present in `EE_GRID_HELPER`, based on the
value of `m_enableGrid`, was not taking into account that another check,
but for the global KiCad setting, would be performed in the `Align()`
method, which is inherited from `GRID_HELPER`.
I've modified all tests of `m_enableGrid` value to also take the global
setting into account, and moved checks for it to `Align()` method, as
it's more consistent this way.
I've also removed an override of the `Align()` method in the
`PCB_GRID_HELPER` class, as it was made redundant by my changes.
Fixes https://gitlab.com/kicad/code/kicad/issues/7402
We had been using each 4th column of spaces. But we needed an absolute
spacing rather than relative to the existing text to match spacing
between the input text box and our stroke font
Can happen if someone hammers away on the probe and/or tune tools
without ever mousing over the schematic in between.
*May* fix the below bug.
Fixes https://gitlab.com/kicad/code/kicad/issues/6876
The variable `m_Initialized` in `BASE_SCREEN` is used by
`SCH_EDIT_FRAME` to mark whether a screen had its zoom level initialized
by the "zoom to fit screen" action. When this variable is `false`, the
function `SCH_EDIT_FRAME::DisplayCurrentSheet()` performs "zoom to fit
screen", modifying the zoom level. This function is indirectly called in
the undo routines, so if `m_Initialized` is not set to `true`, a zoom
change will occur when the user undoes an operation, a behavior that is
undesired.
`m_Initialized` was not initialized to `true` for the null schematic
(the schematic that is loaded if no project is loaded), causing the
aforementioned undesired behavior.
To prevent this, I've changed the `SCH_EDIT_FRAME` constructor to set
`m_Initialized` to `true`, since it zooms to fit screen already. I've
moved `m_Initialized` from `BASE_SCREEN` to `SCH_SCREEN`, as it is used
only in Eeschema, and renamed it to `m_zoomInitialized`, a name I
believe that better describes what this variable does.
I've also introduced the function `SCH_EDIT_FRAME::initScreenZoom()` to
group the "zoom to fit screen" action with setting `m_Initialized` to
`true`, as they often should occur together.
I'd also like to say that I'm not confident whether
`SCH_EDIT_FRAME::DisplayCurrentSheet()` should perform the zoom level
initialization at this point, but I have decided to not change this
behavior for now, as the commit history suggests it's several years old.
Fixes https://gitlab.com/kicad/code/kicad/issues/7343
New strategy isolates all draw/don't draw decisions to the ViewGetLOD
routines, and all dimmed/not dimmed to PCB_RENDER_SETTINGS::GetColor.
The actual drawing in PCB_PAINTER is more-or-less conditon free.
Also adds new layers for pad and via hole walls so that they can be
controlled for high-contrast mode.
Also changes the drawing paradigm so that the pads are drawn even when
not on the high contrast layer, just in low contrast. The hole wall
is drawn in high contrast. This actually makes things clearer to the
user (although to be honest was done to keep from having to re-render
pads when the high contrast layer changes since we have two separate
layers now that we can adjut colours on).
Fixes https://gitlab.com/kicad/code/kicad/issues/7328
They are used for chamfered round rect pads, and can be used for custom shaped pads.
No actual change currently, but the shape rotation of custom pads and chamfered rr pads
can be now used in gerber plots.
Closing the window while dragging with unsaved changes causes a dialog box, triggering a capture-lost event. Closing the window while dragging without unsaved changes causes an assertion about destroying a window with mouse capture.
Two issues found with the locking system used to prevent access to
stale connectivity data during the zone fill process:
1) a std::mutex has undefined behavior if you try to use it to guard
against access from the same thread. Because of the use of wx event
loops (and coroutines) it is entirely possible, and in some situations
inevitable, that the same thread will try to redraw the ratsnest in the
middle of zone refilling.
2) The mutex was only guarding the ZONE_FILLER::Fill method, but the callers
of that method also do connectivity updates as part of the COMMIT::Push.
Redrawing the ratsnest after the Fill but before the Push will result in
stale connectivity pointers to zone filled areas.
Fixed (1) by switching to a trivial spinlock implementation. Spinlocks would
generally not be desirable if the contention for the connectivity data crossed
thread boundaries, but at the moment I believe it's guaranteed that the reads
and writes to connectivity that are guarded by this lock happen from the main
UI thread. The writes are also quite rare compared to reads, and reads are
generally fast, so I'm not really worried about the UI thread spinning for any
real amount of time.
Fixed (2) by moving the locking location up to the call sites of
ZONE_FILLER::Fill.
This issue was quite difficult to reproduce, but I found a fairly reliable way:
It only happens (for me) on Windows, MSYS2 build, with wxWidgets 3.0
It also only happens if I restrict PcbNew to use 2 CPU cores.
With those conditions, I can reproduce the issue described in #6471 by
repeatedly editing a zone properties and changing its net. The crash is
especially easy to trigger if you press some keys (such as 'e' for edit)
while the progress dialog is displayed. It's easiest to do this in a debug
build as the slower KiCad is running, the bigger the window is to trigger this
bug.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/6471
Fixes https://gitlab.com/kicad/code/kicad/-/issues/7048
Many, many KIDIALOGs use OK/Cancel and then rename both buttons to
confirm or deny some action. In those cases we *do* want to store
the deny actions if they check "Do Not Show Again".
Fixes https://gitlab.com/kicad/code/kicad/issues/6979
These dialogs don't have growable features but do have conditional
features, so the size is better adjusted automatically by the code
then by the user.
So many things can go wrong with this control in GTK. We have to
collapse the tree when updating the search string to avoid a crash when
referencing a child object but collapsing the tree will iterate over
elements and crash when we have deleted a symbol.
The temporary fix for this nonsense is to carefully order the calls.
We only need to collapse the search tree if we are not keeping our state
(in other words if we are fully re-building the tree)
Fixes https://gitlab.com/kicad/code/kicad/issues/6910
Versions specific variables that may point to objects that change
through versions, allowing multiple KiCad versions to operate correctly
(even on MSW) on the same machine.
Jeff Young
Tomasz Wlostowski
Wayne Stambaugh
Roberto Fernandez Bautista
Seth Hillbrand
Ian McInerney
Mikolaj Wielgus
Mike Williams
Jon Evans
jean-pierre charras
Marek Roszko
qu1ck
Johan Grip
Alexander Guy
Pradeepa Senanayake
luz paz
david-beinder
luz paz
Johannes Maibaum
Hajo Nils Krabbenhöft
Jonathan Haas
Marcus A. Romer
Fabien Corona
Dominik Wernberger
Zoltan Gyarmati
mitxela