Needed to find bottlenecks in fns, so break out individual sections of
the massive function for easier understanding.
buildItemSubgraphs (one section of the previous function) would build
millions of connections that were never used as stacked pins created X!
connections. Also tested using sets instead of lists and keeping unique
lists to avoid flagging but none of these were as performant as using
flags to remember which items had already been processed.
Fixes https://gitlab.com/kicad/code/kicad/issues/10974
The previous solution cleared all of the fields and force the unit setting
to unit one which would have less than desirable effects when reusing a
schematic in another project. This solution forces all of the instance
settings to the first sheet instance of the current project to prevent file
change churn when changing shared sheets in the current project.
This will not prevent shared schematic file changes when the first sheet
instance data differs between projects.
(cherry picked from commit dd1dbb81f6)
Double-calls are no longer needed when using settings. They can
interfere depending on how the system executes the commands.
Removing keeps the last used plugin as the default, rather than the last
selected
Fixes https://gitlab.com/kicad/code/kicad/issues/11094
Adding a page in the ctor can generate events that change the default
page. Instead, we set the default after all pages have been added,
avoiding the reset
Fixes https://gitlab.com/kicad/code/kicad/issues/10910
The comment regarding bitmaps was true for one more month in 2018, but
hasn't been true for the last 3 years, and clearing the cache is a huge
performance hit.
(cherry picked from commit d58427a519)
Arcs can be altered by the process of changing from on-disk
representation to in-memory representation. Saving back to disk without
modifying the arc should not modify the calculated values.
This stores a copy of the on-disk representation that is only used to
save back to disk in the event that the arc is not modified during
editing.
Fixes https://gitlab.com/kicad/code/kicad/issues/10442
(cherry picked from commit cd7141fd10)
Avoid excess writes to the project directory to dodge issues with remote
file systems not fully writing data.
wxRename works across disk boundaries (in theory) and falls back to
wxCopy/wxRemove when it fails
Fixes https://gitlab.com/kicad/code/kicad/issues/10747
Our heuristics for determining when to append to the undo aren't
sufficient for the two segments of a broken wire. One is new and the
other is changed, but the drag should be appended to the current undo.
Fixes: https://gitlab.com/kicad/code/kicad/-/issues/10865
MapAngles will adjust by small amounts the angles t1 and t2. This
adjustment will determine the CW/CCW drawing. Therefore, the MapAngles
needs to be called explicitly before the Normalize180() call.
Left-to-Right evaluation is not a given for MSVC
Fixes https://gitlab.com/kicad/code/kicad/issues/11050
(cherry picked from commit c9fb95b888)
Before they were 1/2 treated as flags and 1/2 treated as a mode enum.
The ERC flag relaxes constraints on data that is settable in the
schematic editor.
Fixes https://gitlab.com/kicad/code/kicad/issues/11018
(cherry picked from commit 76535d8572)
When double-clicking to open a symbol, it is no longer selected in the
tree. We need to use the `GetTargetLibId()` to pick the correct symbol
(first selected, then current) when using the duplicate command
Fixes https://gitlab.com/kicad/code/kicad/issues/11034
The fix for https://gitlab.com/kicad/code/kicad/issues/10529 kicked the
subsequent IDs for unit selection off the maximum ID value for popup
menus. This increases the maximum popup number (integers are free)
Removes the ability to hide symbol units in file formats. This was
mostly the case for a while but there remained corner cases where we
left a default/hidden option
Fixes https://gitlab.com/kicad/code/kicad/issues/10888
(cherry picked from commit 8c36ba3653)
When reading sheet fields, we need to maintain a consistent order for
the mandatory fields as they are dereferenced by place. We force this
during the `SetFields()` call.
We need to keep the screen blank while initially loading a sheet before
the hierarchy is developed. Avoid resolving text variables during this
process
Fixes https://gitlab.com/kicad/code/kicad/issues/11000
Also makes sure the progress dialog is closed when we're done reading
symbols (it used to stay up for much of the symbol editor initialization).
Also makes sure that any cancel in the preLoad step is honoured in the
sync step. (The preload is done because it is multi-threaded and therefore
faster than the single-threaded sync.)
Also makes sure that individual threads pay attention to the cancellation,
not just the GUI thread.
Fixes https://gitlab.com/kicad/code/kicad/issues/8372
(cherry picked from commit 1f16092e29)
KiCad has stricter requirements for what consititutes an annotated
symbol. This checks for and corrects annotations that begin with '#',
do not begin with a non-numeric character or do not end with a number
(cherry picked from commit 78620347c8)
Fixes https://gitlab.com/kicad/code/kicad/issues/10760
Center rotation will often misalign lines to the grid, which is bad
until fix off grid items on the schematic.
We can rotate a connection end into a new connection or to become
collinear with an existing line, so we need to check for this at the end
of rotation.
Fixes: https://gitlab.com/kicad/code/kicad/-/issues/10565
If a wire intersects more than 2 pins from the same symbol, we don't
trim connections. If the wire intersects exactly two pins from a single
symbol, we remove the wire between the pins.
We avoided doing this before because we didn't have a good metric for
trimming vs. not. But the per-symbol calculation feels like the least
surprising option.
Fixes https://gitlab.com/kicad/code/kicad/issues/10909
Fixes https://gitlab.com/kicad/code/kicad/issues/1857
Apparently at some point during V5 development, symbol cache library names
were saved by replacing the LIB_ID separator character ':' with '_'. This
caused the cache look up to fail there by skipping the symbol rescue which
could result in broken schematics.
I have no idea where this happened during V5 development. The video demo
in the HEAD of the 5.1 branch shows the issue. All of the other demo cache
libraries are correct.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/10488
gcc doesn't need it and computes during compile regardless, but experimenting in godbolt, both clang and MSVC actually do need it or else it's partially computed at runtime
(cherry picked from commit ca7840334c)
We need the item to be currently selected when repeating the last symbol
insert. This ensures we have a valid selection before starting the move
tool and possibly getting odd system behavior from mouse positions
Fixes https://gitlab.com/kicad/code/kicad/issues/10619
We can't stay in a sheet that is deleted by the action of undo. Checks
the current sheet and, if we are currently using it, ensures that the
following action is to leave the current sheet
Fixes https://gitlab.com/kicad/code/kicad/issues/10733
When the grid editor is shown, we also want to ensure that the grid gets
resized to account for the new editor. Queuing the event will ensure
that happens after the editor is emplaced in the grid
Fixes https://gitlab.com/kicad/code/kicad/issues/5690
On GTK, the full editor needs to be shown and, for some systems, the
editor will not show correctly. By yielding we should allow the time to
recalculate sizes for the editor in the grid
Fixes https://gitlab.com/kicad/code/kicad/issues/5690
Our undo/redo list is tied to the screen that gets cleared when we
reload the file after renaming. We need to clear the Undo list at this
point to prevent possible crashes when moving too far back in the stack
Fixes https://gitlab.com/kicad/code/kicad/issues/10504
Eeschema/libedit support open, filled polylines. When printing, this
should be drawn as a fill, followed by the polyline to avoid assumptions
about closed polygons in the graphic code
Fixes https://gitlab.com/kicad/code/kicad/issues/10537
Rather than skipping our endEdit() call in libedit, we need to
parameterize whether we want the shape open/closed. Closed will remove
the last segment if it lands on the first point. We don't want that
but we do want to remove the last point if it duplicates the second to
last (in the case of double-clicking)
Fixes https://gitlab.com/kicad/code/kicad/issues/10334
SHAPE_LINE_CHAIN will, by default, remove duplicate points when
appending. We don't want to do this when constructing our polygons in
symbol editor, so we need to explicitly call the routine without
deduplication
Fixes https://gitlab.com/kicad/code/kicad/issues/10289
When leaving the main point editor loop, we need to ensure that the edit
point is cleared as well. This is usually cleared by a ClearEvent call
but in case this is missed, we need to ensure that other tools are not
caught out
Fixes https://gitlab.com/kicad/code/kicad/issues/10288
Undo/Redo can change the symbol referenced by the symbol drawing tool.
This updates the current symbol before important steps to keep the
system synced
Fixes https://gitlab.com/kicad/code/kicad/issues/10125
Exporting the Bill of Materials uses the netlist exporter. Both of
these routines were stripping the extension, leading to projects like
"test.project.kicad_pro" having the project name stripped. We separate
the BOM netlist export from generic plugin netlist export to allow the
correct behavior when exporting netlists and generating BOMs
Fixes https://gitlab.com/kicad/code/kicad/issues/10270
We were only checking for the selection of a single line that gets
merged. This prevents the merged line from being highlighted but not
selected after merging
Fixes https://gitlab.com/kicad/code/kicad/issues/10202
This is particularly important when reading legacy projects as 0 is
what signals the painter to use the old algorithm (1/2 pin name/number
size).
Fixes https://gitlab.com/kicad/code/kicad/issues/10100
(cherry picked from commit f089cf44a7)
It is no longer acceptable to set a LIB_FIELD index to -1. This was
trashing the sorting on load cause all kinds of false field comparison
failures. There are now assertions when attempting to use -1 as a field
index.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/9811
Always use full sheet paths when storing sheet instances. Partial sheet
paths cannot be full resolved resulting in lost page numbers when renaming
sheet file name.
Fixes#9782
And undo the previous changes in SCH_TEXT::Rotate().
We continue hitting this with a wrench until it gets fixed or becomes
broken beyond any recognition.
Wayne Stambaugh
Seth Hillbrand
jean-pierre charras
Jeff Young
Mike Williams
Steffen Mauch
Marek Roszko
dana
Roberto Fernandez Bautista
Jon Evans
dsa-t
Mikolaj Wielgus
Adam Wolf
Franck Bourdonnec
kliment