* Move column width manipulation into the lib tree adapter
* Fix issue with GTK where if someone types too fast, the part
column gets 0 width.
Fixes: lp:1841584
* https://bugs.launchpad.net/kicad/+bug/1841584
There is a rendering bug in GTK3, which appears to be
an upstream GTK issue.
This can be worked around by, when filtering, ensuring the
*parent* item of the selected item is visible. This will
not cause the selected item to not be visible, as the selected
item will be the first shown child. So it will be visible, as long
as the list box is greater than a single row high, which it will be
in all practical scenarios.
This is done on all platforms, as it has a beneficial side-effect:
the parent library of the selection is naturally shown to the
user, so they don't need to scroll up to see what library their
current filter selection was in.
Fixes: lp:1804400
* https://bugs.launchpad.net/kicad/+bug/1804400
The indent size was estimated by the width of characters. But MSW
indent is substantially different from OSX and Linux. It also cuts off
the middle characters rather than the end leading to poor display if the
width does not fit. This uses the system setting for indent to account
for the indent spacing + 'M' to account for the arrow inset.
Fixes: lp:1815401
* https://bugs.launchpad.net/kicad/+bug/1815401
When filtering, we update the width of the displayed column to ensure
the full text is visible to the user. Check is rough, based on line
width (doesn't completely account for differing char widths) but is
sufficient for the approximate difference
Fixes: lp:1815401
* https://bugs.launchpad.net/kicad/+bug/1815401
Fixes: lp:1788495
* https://bugs.launchpad.net/kicad/+bug/1788495
When the user cancels the footprint load, we should assume they are
canceling the placement of a new footprint. This also adds sanity check
when populating the list
Fixes: lp:1814181
* https://bugs.launchpad.net/kicad/+bug/1814181
Be more intelligent about sorting lib tree items. (Footprint
entries, for instance, come out of an already-sorted list.)
Don't recreate menus twice when laoding Footprint Editor.
More pervasive use of WX_FILENAME to avoid expensive calls to
wxFileName::SplitPath() and string concatenation.
For POSIX kernels do all the work on the file-system side so we
don't have to keep converting back and forth between encodings.
1) don't call UI-level LIB_PIN routines when reading library --
not only are they a performance hit, they set the modified flag
too
2) limit progress dialog updating to 15 times a second (this
had crept back up to 31% of the time spent loading libraries)
Fixes: lp:1734773
* https://bugs.launchpad.net/kicad/+bug/1734773
Apparently explicitly requesting a column to be sorted breaks component
filtering on Windows. Without SetSortOrder() call the list is sorted and
filtering still works as expected.
Fixes: lp:1739412
* https://bugs.launchpad.net/kicad/+bug/1739412
Please never ask me about the SetSortOrder() part. When it is set to
'true', wxDataViewCtrl::EnsureVisible() does not always work correctly,
so sometimes the best search result is not shown in the window (no
pattern found). Now it is set to 'false' to avoid the described
behavior, but the sorting order is reversed on Windows, not on Linux.
I could not test it on OSX, but I am sure it paints walls in your room
pink. /me burst into tears.
Fixes: lp:1738696
* https://bugs.launchpad.net/kicad/+bug/1738696
It turns out the fastest way to update wxDataViewCtrl is to nuke the
model and rebuild it from scratch rather than via
ItemAdded()/ItemDeleted() update notifications.
Fixes: lp:1734773
* https://bugs.launchpad.net/kicad/+bug/1734773
Tree nodes now have an additional field 'InTree' to determine
if the view is aware of its existence. This way, there is no need
to rebuild the data structures from scratch when they need to be
filtered.
Add missing CMP_TREE_NODE_LIB_ID type that was inadvertently deleted by
the symbol library table implementation that broke the preselected symbol
tree expansion and highlighting.
Changed CMP_TREE_NODE_LIB_ID type from ALIAS to LIBID for more accurate
description.
Use LIB_ID in COMPONENT_SELECTION object instead of strings of the
library and symbol names.
Change all of the component tree helper objects to use LIB_IDs instead of
LIB_ALIAS pointers. LIB_ALIAS pointers are dangerous to use because they
can be deleted in the symbol library editor while the component chooser
dialog has copies of them. With LIB_IDs, the LIB_ALIAS pointer is found
on demand and can be guaranteed to be valid.
Update the chooser dialog to load the symbol library table instead of the
libraries defined in the project file and return a LIB_ID instead of a
LIB_ALIAS pointer.
Modify SCH_BASE_FRAME::SelectComponentFromLibrary() to handle the LIB_IDs
returned from the component chooser dialog.
- In the listing, display alias names in italics
- In the info panel, display per-alias description correctly, as well as
root description
Fixes: lp:1676190
* https://bugs.launchpad.net/kicad/+bug/1676190
- Stop wx from sorting the items. We've already sorted them, and
profiling shows our sort is faster than theirs.
- Go back to using strings as cache keys, because using pointers means
the cache doesn't survive from one invocation to the next. Switch to
wxHashMap because it can use wxString keys in an unordered (faster)
map.
This refactors COMPONENT_TREE_SEARCH_CONTAINER into a Model-View-Adapter
architecture comprising:
- eeschema/cmp_tree_model.h
- CMP_TREE_NODE: Base class representing a searchable library
set with scoring and sorting
- CMP_TREE_NODE_UNIT
- CMP_TREE_NODE_ALIAS
- CMP_TREE_NODE_LIB
- CMP_TREE_NODE_ROOT
- eeschema/cmp_tree_model_adapter.h
- CMP_TREE_MODEL_ADAPTER: mediator between wxDataViewCtrl (via
wxDataViewModel) and CMP_TREE_NODE*
+---+ +------------------+
+---+ Generates | A | | VIEW |
| M | from libs | D | wxDataViewModel |------------------|
| O | <---------- | A | <------------------> | wxDataViewCtrl |
| D | | P | |------------------|
| E | <---------> | T | <------------------- | wxTextCtrl |
| L | UpdateScore | E | UpdateSearchString() |------------------|
+---+ | R | | |
+---+ +------------------+
Representing the data with a proper model allows the wxDataViewCtrl to
be updated in bulk, which is significantly faster than the old method of
populating it one item at a time. This also adds flexibility if more
data is to be added in the future (which may come in handy with the
upcoming .sweet format, as that adds more ways components can be related
to each other).
Replacing the wxTreeListCtrl with a wxDataViewCtrl also significantly
reduces the general GUI quirkiness, as wxDataViewCtrl is much more well
behaved.
Ian McInerney
Jeff Young
John Beard
jean-pierre charras
Seth Hillbrand
Maciej Suminski
Camille
Wayne Stambaugh
Simon Richter
Oliver
Chris Pavlina