Importing Eagle files allowed for invalid zone values for clearance and
minimum width that triggered asserts in the 3d-viewer.
Fixes: lp:1801188
* https://bugs.launchpad.net/kicad/+bug/1801188
Fedora packaging errors with ambiguous python shebang since current
rawhide (30).
So we should remove it as the footprint wizards are not to be executed
standalone.
Variable netBuffer contains all NET_INFO items, whereas max_netcode
value is determined only basing on pad nets. After a netlist update
there still might be a NETINFO_ITEM that is not assigned to any pad and
has a net code higher than calculated max_netcode. Such situation
results in a out-of-bounds access in sortNetsByNodes().
Separate out zone edges (which are quite specific, and should
win over most other things) from zone fills (which are pretty
generic and should only be considered when there's nothing else).
Also excludes zone fills when the zones aren't being shown as
filled.
Display GALs had an incorrect world unit value set. Now the world unit
value says how many internal units are in an inch, in accord with the comments
in the GAL header. Bitmap drawing code relied on the information about DPI,
so scaling worked differently for display and print GALs.
It is possible for an async thread to simply be deferred if not
specifically set to async. In which case, until the return value is
requested, the loop may simply wait.
By decoupling from std::thread, we can avoid the wait/sleep cycle in
checking the std::atomic completion variable. The std::future variables
are immediately returning without the additional atomic check cycle.
This minimizes the thread usage, particularly during online actions such
as moving items and routing.
UpdateRatsnest: Minimized the number of atomics by utilizing future
promise. Avoided updating node-less nets. Node-less nets are common in
the dynamic ratsnest because the net vector is created from the board
nets by value but only populated with a subset. So the dynamic ratsnest
calculation always saw every net as dirty and spun up extra threads.
searchConnections: ParallelThreadCount is no longer lower bounded.
Small item counts are handled inline rather than with a separate thread.
When processing the selection filter, items were deselected before being
reselected after passing through the filter. This adjusts the logic to
only deselect those items that are filtered out.
Some windows machines seem prone to sleeping for excessive times. When
called in the main thread, this may prevent the system from responding
to Windows 'Are you alive?' calls.
TF.FileFunction,Copper,L1,Top,Signal is now TF.FileFunction,Copper,L1,Top
This param is not useful, and probably never correctly set by users, because is is not used by Pcbnew.
(It is used only for dsn export)
The implementation relied on the fact that PyString_AsString is returning NULL
for invalid inputs (and causing an error condition), which is then fixed by the
call to FROM_UTF8. Now, we will simply return an empty string if Py_None is
returned by a python fuction, which resemebles the old behaviour of those methods
Rather than forcing each thread to iterate over the full nets list to
find the dirty items, we collect the dirty items first before asking the
threads to address them. This prevents thread spinning. We also
amortize the cost of calculating on an 8-net per thread basis rather
than 2.
Locks protect the std::set in each item. Devolving the mutex to the
CN_ITEM allows multiple threads to make simultaneous connections to
different items where they do not conflict.
The connectivity files were unwieldy. This separates them logically
into data, algo and items where the items classes are those that hold,
surprise, surprise, the items, lists and clusters.
Refactor the polygon event loop into its parent function to allow access
to the params structure. Active layer now updates the drawing layer for
add/cutout zones.
Fixes: lp:1797483
* https://bugs.launchpad.net/kicad/+bug/1797483
Dynamic casts should only be used when we explicitly check for the
resulting pointer to be NULL. Where we know the class is castable we
can use static_cast, save on overhead and ensure our resulting pointer
is non-null.
When using the board edit, align tool automatically moved footprints
instead of pads but distribute tool did not. This standardizes the
behavior.
Fixes: lp:1796997
* https://bugs.launchpad.net/kicad/+bug/1796997
The evaluation of count_if resulted in a 0 or 1, leading to segment zone
fills always running on single processor and the progress bar showing
incorrect status. Likely infrequently hit as only affected segment
fills.
Block rotate and mirror now work like they do in Pcbnew (that is
they're treated as part of the block move/duplicate/whatever
rather than finishing it).
Fixes: lp:1780794
* https://bugs.launchpad.net/kicad/+bug/1780794
If the user is carefully selecting the zone edge, we do not filter the
zone preemptively from our disambiguation list in the same manner as we
do for selecting on zone copper.
Fixes: lp:1796719
* https://bugs.launchpad.net/kicad/+bug/1796719
The board design settings file keeps our default values for various
thicknesses. We use this to replace a number of magic numbers with the
defined value to help self-document the code.
The check against S_POLYGON was always false because we set the default
type to S_SEGMENT. This unifies the magic numbers to the board design
settings file
Disabling the grid should leave the snapping in place. But our disabler
returned the current point as closest, effectively always ensuring the
cursor position was the closest snap. This only uses grid when
requested.
This is temporary until we figure out a way to do multiple constraint
handling in GRID_HELPER or the EC_* class. This will work as long as
the EC constraint helper places the endpoint on either the primary or
aux grid.
Fixes: lp:1796540
* https://bugs.launchpad.net/kicad/+bug/1796540
We no longer require this as we calculate the number of segments based
on the maximum deviation rather than fixed numbers.
Fixes: lp:1780518
* https://bugs.launchpad.net/kicad/+bug/1780518
Get the board settings when changing the drawing layer. These include
default line thickness that should be updated.
Fixes: lp:1796489
* https://bugs.launchpad.net/kicad/+bug/1796489
When drawing the crosshair should track the element while the mouse
provides the position to calculate the next snap.
Fixes: lp:1796524
* https://bugs.launchpad.net/kicad/+bug/1796524
The dimension tool is used to measure physical dimensions on the board
therefore it should be able to snap to any other object, not just those
on its own layer set as is the case for other items.
This also deals with a corner snap case for constraints by first
snapping, then constraining, then aligning to grid. This ensures OoO
for alignment from least to most constraining.
When considering potential targets for snap points, we need to exclude
the currently selected item as this will mean that we always snap to
ourselves when editing an existing item.
Fixes: lp:1796507
* https://bugs.launchpad.net/kicad/+bug/1796507
When drawing segments, the segment will be NULL'd after committing, at
which point we lost our snap layer. Choosing snap layers from the
current drawing layer is a more robust method.
Fixes: lp:1796467
* https://bugs.launchpad.net/kicad/+bug/1796467
Deselection by clicking on a zone makes working with plane zones easier
by allowing the selected item to de-selected by clicking on a zone.
Once the selection is cleared, the zone may be re-selected as normal.
Clicking on non-zone items while a separate item is selected will still
select the new non-zone item without requiring a second click.
Updating connectivity should not need to iterate over the full item list
in each thread. Instead, we collect the dirty items first and then
iterate only over the dirty list.
Our connectivity data threads the ratsnest update based on the number of
nets that have changed. If we are only changing one net (the case when
routing) then making a full thread to update the net is pointless
because we wait for the thread to finish before continuing. This spins
that off to an inline lambda for single updates while still making
threads for multiple ratsnest changes.
Seth Hillbrand
jean-pierre charras
Maciej Suminski
Jeff Young
Nick Østergaard
Thomas Pointhuber
Wayne Stambaugh
Roman Beranek
Tomasz Włostowski