1) If "support on build plate only" is enabled, the support columns are
newly trimmed to not land on top of an object. However this may make
the column too small to be stable.
2) Support enforcers newly take precedence over "supports on build plate only"
and over "don't support bridges".
3) Some refactoring of the support generator code for clarity: Reduced
some of the worst spagetti offenders.
Fixes Support generated even if support on build only activated #915
Fixes Bug: supports on build plate only #1340
Fixes Bottom interface layer is not generated , support on build plate only. (long open defect) #4199
Fixes option "supports on build plate only" does not work #3980
Fixes No support interface layers generated #1997
Fixes Feature Request: Option to combine results of 'support from build plate only' and 'support enforcers only' #2801
Fixes Support interface isn't generated: build plate only + blocked by model + support enforcer #3831
Fixes Support Enforcer don't create interface layers #5748
Fixes Support Enforcers Don't Have Top Loops/Raft #1870
Fixes Don't cancel support enforcers with "don't support bridges" #5105
Problem description:
When "Print Settings -> Multiple Extruders -> No sparse layer" is enabled, then "Smart" Wipe Tower is used for wiping.
As a result, each layer with tool changes is splited for min 3 parts: first tool, wiping, second tool ...
But vertical slider wasn't respect to this case.
were missing abruptly when going down.
The issue was caused by extracting support areas from a grid and
filtering the extracted islands by intersection with the input islands.
Sometimes the input islands were a bit bigger than the extracted contour,
thus some of the samples of the input islands did not fall into
the extracted contour.
the Chromebooks share their file system to Linux using the 9p file
system, which does not support setting file ownership. Newly PrusaSlicer
will detect platform and it will not panick if copy_file() cannot set
file ownership after copying. It just logs the incident, and on
chromebooks the loglevel for that incident is "Info", not "Error".
Adjusted the full screen mode to contain menu bar.
Moved Platform.cpp/hpp to libslic3r
* GUI_ObjectList code refactoring:
The MenuFactory structure contains functions related to the context menu and bitmaps used to different volume types.
The SettingsFactory structure contains functions to getting overridden options, its bundles and bitmaps used to setting categories.
Fixed bugs/crashes:
1. Add object -> Add Settings from 3D scene -> Right click on object => Part's Settings list instead of object's
(Same behavior if something else but Object is selected in ObjectList)
2. Add settings to the part -> Change part type to the "Support Blocker/Enforcer" -> Settings disappears (it's OK) but =>
Save Project -> Open project => Support Blocker/Enforcer has a settings
3. Add part for object -> Change type of part -> Change monitor DPI -> old type icon appears
4. Select all instances in ObjectList -> Context menu in 3D scene -> Add Settings -> Select some category -> Crash
* ObjectLayers: Fixed a crash on re-scaling, when some layer range is selected
* Fixed OSX build
* Added menu item "Split to Objects" for multipart objects
+ Fixed bug: Add 2 parts,
Add some settings for one part
Delete part without settings => Single part object without settings, but settings are applied for the object.
+ Next refactoring: use same menu for Plater and ObjectList
Namely, on ChromeOS virgl flips red/blue channels at least on some computers with multi-sampling enabled.
It seems it is sufficient to disable multi-sampling after the OpenGL context is created.
