Internally time stamps are defined as unsigned long. On some platforms,
time_t is defined as unsigned long long. Cast all time_t instances to
unsigned long to prevent the compiler from complaining.
*) Extend SWIG support deeper into the BOARD class.
*) Move swig *.i files into a directory identified for SWIG, in preparation
for a parallel universe involving Phoenix's SIP.
*) Move swig files which will be common to both eeschema and pcbnew into
common/swig.
*) Sketch out a "common" python module, and plan on dovetailing that into a
libkicad_shared.{dll,so}
*) Add common/swig/ki_exceptions.i and define a macro HANDLE_EXCEPTIONS()
which is to be applied to any function which needs C++ to python
exception translation.
*) Move the test for SWIG tool into top level CMakeLists.txt file for use
in all python modules beyond pcbnew, i.e. eeschema and common.
*) Add SWIG_MODULE_pcbnew_EXTRA_DEPS which generates a better Makefile, one
which rebuilds the swig generated *.cxx file when one of its dependencies
change.
*) Re-architect the board.i file so that it can be split into multiple *.i
files easily.
*) Make some KIWAY from python progress, in preparation for Modular KiCad
phase III.
This option is not yet activated because the net attributes are not yet fully fixed by Ucamco, in Gerber file format specifications.
(To activate it, see dialog_plot.cpp, line 43)
add overloads for NODE::Add( ... ) taking pointers to specific item types (retain old private add-Functions, they will come in handy later)
LINE overloads now take by reference, to document their special treatment.
updated code throughout affected by these changes
The DRC also runs per-segment tests, not just segment-segment, so the last
segment needs to be tested as well, with an empty list of segments to test
against.
* When Pcbnew is launched in the stand alone mode with no board file command
line option, the project path is set to the current working directory. The
user may not have write access to the current working directory which would
cause the auto save to generate an error. If the user does not have write
access to the current working directory, set the path to a platform specific
temporary path that the user has write permission.
* When Pcbnew is launched in the stand alone mode with no board file command
line option, the default BOARD object has no file name. Set the file name
to "noname.kicad_pcb" to fix incorrect auto save file name and set the path
to the user's platform specific document folder.
* Delete orphaned auto save files when closing without saving due to broken
auto save file generation logic which prepended the file name with "$" rather
than "_autosave-".
* Fixes: lp:1596382
* Split and rewrite the preview window and canvas.
* Create a new class for handling the board information.
* Adds new render targets: openGL, legacy, and ray tracing.
* Render targets take full advantage of the new 3D plugins system and 3D cache
for a fast 3D model loading.
* Faster board loading.
* New OpenGL render is faster than the old one.
* New ray tracing render target with a post processing shader.
* Use of new 3D plugins (WRL, X3D, STEP and IGES) and 3D model caching.
* Preview of 3D model while browsing the file name.
* 3D preview of the footprint while adding / align 3D shapes.
* Render of 3D models according to attributes: Normal, Normal+Insert, Virtual.
* Pivot rotation centered in one point over the PCB board.
* Shortcuts keys improved for XYZ orientation..
* Animated camera.
* Switched hashtables.h over to std::undordered_map from boost version.
* Added new macros DECL_VEC_FOR_SWIG() and DECL_MAP_FOR_SWIG() in macros.h.
These along with future DECL_HASH_FOR_SWIG() unify the declaration to swig
and C++ so that the resultant type name is common in both languages, and
the types AGREE.
* Fixed swigging of NETINFO_ITEM and NETINFO_LIST via magic.
* Newly exposed (python wrapped) are: D_PADS, TRACKS (was TRACK_PTRS),
NETNAME_MAP, NETCODE_MAP, wxString (without constructor purposely, read
comment in wx.i), MARKERS, ZONE_CONTAINERS, NETCLASSPTR, KICAD_T types.
* std::vector<SOMETHING*> tends to end up named SOMETHINGS in C++ and python.
Having the name consistent between like types is helpful, and between
languages. std::map<> ends up as SOMETHING_MAP.
* NETINFO_LIST::m_netNames and NETINFO_LIST::m_netCodes are now std::map
instead of hashtables, because swig does not yet support std::unordered_map.
* You can now get to any netclass or net info. NETNAMES_MAP and NETCODES_MAP
are traversable basically the same as a python dictionary using a python
string (not wsString) as the key! The wxString typemap converts python
string to wxString before the lookup happens. Iteration also works.
* Factor out PROPERTIES object from the PCB plugin code and move it into
common so it can be used by both the Pcbnew and Eeschema plugins.
* Add schematic I/O plugin manager for loading and saving schematic and
component library files.
* Add initial attempt at a parser for current schematic file format. This
parser will be infinitely more strict than the current parser which is very
forgiving in what it parses.
* Make minor changes to the base bitmap class to support the new parser.
* Add find root sheet support to sheet object to allow fetching the root
sheet from any sheet in the stack.
It is especially true in some Window Managers like KDE, but can be true in any WM, when initializations are made by TransferDataToWindow() method, called by the default wxInitDialogEvent handler.
This revised nginx.conf holds the cache for one week, truly. For some reason there are
two tuning knobs required to make one adjustment (cache duration). Without this fix the
cache was only held for 10 minutes. The most important change is the "inactive=1w" which
as I implied, is the *second* reference to one week. Sorry for this oversight.
Nginx can be run on something as simple as a raspberry PI. I run it on a network file
server attached to my desktop via copper ethernet. By caching the footprint libraries
from github, I can load all github libraries in about 2.5 seconds, making github the
fastest plugin of any kind.
This feature was not reliably available: neither pl_editor nor GAL supported
it. It has been replaced over the past few commits with a new zoom-to-selection
tool available in all applications and modes.
While defining functions in another namespace is technically allowed as
long as the definition can be matched to a declaration, this can lead to
ambiguous resolutions, such as here (GAL vs KIGFX).
This feature cannot work reliably for reasons explained in lp:1463505. Remove
the option on OS X builds and force cursor to the small one.
https://bugs.launchpad.net/kicad/+bug/1463505
> Bernhard Stegmaier (stegmaier) wrote on 2016-01-14:
>
> As far as I see from the code this is no bug but done on intention.
>
> The reason probably is that on OSX in the legacy canvas you don't have the
> XOR-drawing as on other platforms. To the best of my knowledge, this would
> mean that on every mouse move you would have to repaint the whole window just
> to get rid of the "old" cross.
>
> So, I would say this is a clear "won't fix".
> We could think of just removing/disabling the button in the legacy canvas.
Coverity: CID 147344 Unchecked dynamic_cast
If a file that parses validly in Parse() but does not return a BOARD is loaded,
no exception will be thrown but the dynamic_cast will return NULL. This NULL
is tested by wxASSERT(), which does not actually halt anything (especially in
release builds), and then we proceed to board->SetFileName().
This can be demonstrated quickly by renaming a .kicad_mod to .kicad_pcb and
trying to load it. pcbnew will crash.
Currently, kicad_pcb files have a (version ...) tag, but it is ignored. This
commit does the following:
1. Parse the version. If it's later than the last supported version, present an
alternative error message suggesting an upgrade if the file does not load
correctly. The version will be interpreted as an integer YYYYMMDD to suggest
a KiCad release date.
2. Accept a (version ...) tag also in kicad_mod files, but do not write them
yet. If no version tag is present in these files, assume the version to be that
of the current format version at the time of this commit.
This is meant to be merged to the 4.x stable series, and preps for KiCad 5
which will start emitting version tags also in footprints - users with what
will then be 'old stable' will not get a parse error when we introduce this for
footprints, and we can safely increment the format version later.
It should help if (or when) the internal angle unit used in kicad will be changed from 0.1 degree (a relic of code written for PCs without fpu) to degree ( a more natural unit).
* Always hide footprint text when the layer that the text resides is disabled
in the layer manager. There is no longer a difference between the value and
reference text compared to other field text.
* The usual smattering of coding policy fixes and general code cleaning.
* Paths defined via pcbnew->Preferences->Configure Paths are now included in
the 3D file browser's drop-down list except for KICAD_PTEMPLATES, KIGITHUB,
KISYSMOD, any paths resembling a URL, and any non-existent paths. When an
absolute path is shortened using one of these path aliases, let's say
'KISYS3DMOD' then the name appears in the pcb file as "${KISYS3DMOD}/blah.wrl".
If a user defines and actual environment variable with the same name, then
that user defined variable will have precedence over the internally defined
variable.
* Paths relative to the current project directory are no longer represented
as "some/path/to/file.wrl". For compatibility with previous KiCad versions
it shall always be stored as ${KIPRJMOD}/some/path/to/file.wrl.
* Various restrictions on alias path values have been removed since the
imposed restrictions could result in corrupted user data as aliases are
silently dropped.
* Overall the patch should improve back-compatibility by providing a ${ENV_VAR}
mechanism for supporting different model root directories and ensuring that
paths relative to KIPRJMOD are expressed in a way that is compatible with
earlier KiCad versions. This allows users more flexibility and the ability
to work in a way that ensures 3D model files can be correctly resolved in
earlier versions of KiCad. Users who do not care about back-compatibility
may prefer to use the newer alias system.
Fixa also a corner case which crashes Pcbnew: if the dialog is opened, and the board was changed (reloaded, cleared), pcbnew crashed because many pointers stored in dialog become invalid. Now the dialog is just closed.
- When the Kicad manager is iconized, all other child frames are iconized (normal wxWidgets behavior)
- For viewer frames in modal mode, there is a more annoying issue: they have the wxFLOAT_ON_PARENT frame style on unix, and wxSTAY_ON_TOP on Windows.
wxFLOAT_ON_PARENT frame style is incorrect (although on most WM it works) because the parent frame (Kicad manager Frame) is not the caller. It is usually a main editor frame.
It does not work on Windows (wx STAY_ON_TOP is used).
* Now each editor frame has a null parent.
Therefore iconizing the Kicad manager frame does not iconize other frames.
* Viewer frames have null parent in normal mode and the caller parent in modal mode (therefore wxFLOAT_ON_PARENT frame style is always and correctly used)
* References to opened/closed main frames are no more managed by the (complicated) kicad code. Instead of, a non critical and more easy to understand code just uses FindWindowByName to know if a main window exists or not.
These changes do not fix all issues about Kicad frames brought to the foreground or the background, but it fixes a few issues, and AFAIK do not add other issues.
The attached patch makes the following changes to import DXF:
1. correctly implements scaling based on DXF $INSUNITS -
at least where INSUNITS is sensible. I ignore units like miles,
gigameters, AU, lightyear, parsec ...
2. correctly implements LWPolylines, adding an ARC segment
rather than a line segment when bulge != 0. A number of users
have been complaining about getting the wrong board shape
(chamfered corners rather than rounded corners) on DXF Import,
now they shouldn't complain.
Following rev 6627 (git 2777182), if a non-duplicatable item was the object of
a Duplicate action, the original item would be picked up, potentially
misleading the user into thinking something was actually duplicated.
Do not assert that the item to be duplicated can be duplicated, just don't
duplicate it. The user can trigger a Duplicate on items that aren't
duplicatable, like field texts - he'll figure out that they can't be duplicated
when they aren't, no need to deliver a scary warning and risk an abort.
Do not assert that the item to be duplicated can be duplicated, just don't
duplicate it. The user can trigger a Duplicate on items that aren't
duplicatable, like field texts - he'll figure out that they can't be duplicated
when they aren't, no need to deliver a scary warning and risk an abort.
TODO: Don't pick up the duplicated selection if none of the items in it were
actually duplicated, it's misleading.
* On MinGW64 and possibly other 64-bit platforms, time_t is defined as a long
long unsigned integer but all of the scanf and printf format specifiers used
to save and load the schematic files are %lX which is a long unsigned integer
which causes compile warnings. Add casts and temporary variables to eliminate
the compile type warnings.
I tried to merge some constants used both in legacy mode and gal mode (which were, before this patch, separate constants).
There is still a serious work to avoid different calculation code for the same text in draw and plot functions.
Work in progress to merge these calculation functions.
* No initial copied object changed (this was a serious bug to modify these objects. Previous version modified references and other texts using a very stupid algorithm).
It also fixes bug 1549231
* only new pads are numbered (therefore renumbering is used only in footprint editor)
* remove not working and useless feature in circular array: now only use number for pads (others options using alphabetical letters are removed: did not work corectly, and were useless)
* a more clear option is used to choose if the pads are numbered from a choosen value, or from the first avaible value
* Adding a warning messsage if a parameter is incorrect.
* Move add_directory()s in main CMakeList.txt after all find_package()
macros are run so all flags in the main CMakeList.txt are propagated
to the sub-folder CMakeList.txt files.
* Remove commented out include of config.h
* Make the link maps only build on linux as -${TO_LINKER},-cref
-${TO_LINKER},-Map=blah just gives warnings on osx w/ clang
* Make the link maps only build with flag -DKICAD_MAKE_LINK_MAPS is
defined during CMake configuration as they are highly specific.
* Moved the code for setting link maps into the main CMakeList.txt
file to avoid duplication.
* Removed -D__ASSERTMACRO__ from osx as its no longer needed
* Removed a couple of other OSX only things that wouldn't work anyway
* Moved set (BU_CHMOD_BUNDLE_ITEMS) to the main CMakeList.txt as
otherwise it would only work for the KiCad build not the other
applications
* Made KICAD_SCRIPTING_MODULES enable KICAD_SCRIPTING as currently if
you have modules enabled without the scripting base it will on build.
This could be changed to a fatal error saying you need to also enable
scripting but it seems unnecessary.
* Removed duplication of pcbnew.py install code under modules and
scripting since you can't have modules without scripting now
* File import: use case insensitive comparison to choose import plugin, from the file extension (It is not always in lower cases, especially on Windows.)
* onleftclick.cpp: fis a minor issue, which happens after switching form GAL to Legacy canvas: sometimes a board item is left selected from GAL, and creates false warning message in legacy.
* Dialog copper zones: Always enable thermal shape settings, because even with no thermal, some pads can use thermal option in local pad settings.
* GAL mode: shows page limits in gray color, like in legacy mode, and according to comments in sources.
* Fix a StrPrintf() %s count mismatch bug
* Better comments
* Use std::string::data() instead of a ref to the first char to get the str::string buffer (should not change anything with GNU lib)
* The resource setting for bitmap2component is too late in the CMakeLists.txt,
and is being ignored. Bitmap2component does not have an icon resource on
Windows. Moving the entire section resolves the issue.
* The other programs only have the mingw special case listed, not the generic
declaration for the resource file so added them.
The standard library requires iterators passed to functions that modify the
container to be mutable iterators, but GCC's implementation accepts
const_iterator in some places where these are only used to mark a place,
but the actual modification happens through a different parameter.
As this breaks implementations that use the passed iterator to modify the
container (e.g. because they use a different data organization), this is
not portable; because we already have a non-const reference to the
container anyway, this is trivially fixed as well.
In the C++ standard, this function is only defined for floating point
types, and integers cannot be implicitly converted. Using explicit
conversions avoids a GCC specific extension to the standard library.
This was apparently left in from debugging earlier, and should no longer be
needed. Since it uses a GCC extension, it makes compilation on others fail.
In the autorouter code, the value 0x80 is assigned to MATRIX_CELL, which is
an overflow for a signed 8-bit type. As this type is used as a bit mask,
there is no point in having a sign bit anyway.
The C++ preprocessor is actually not required to process "true" and "false"
correctly. This works in C if <stdbool.h> is included, because these are
then macros themselves, and resolved correctly, but C++ requires them to be
keywords, so no such macros exist, and the preprocessor can treat both as
undefined/zero.
Mostly cosmetic change, although there are compilers that choke on this.
The C++ standard specifies that classes contain themselves as members,
probably so they shadow any other definition of the same name for their own
member functions, but there is really no reason why the class name should
be duplicated here.
* Prevent libcurl from inadvertently being initialized twice.
* Dynamically load libcurl only when required.
* Reduce the number of worker threads when loading footprint libraries to
prevent issues with the GitHub plugin.
This simplifies the pad DRC significantly and fixes a bug when two pads
are close to each other in a right angle, when the edge distances are
below the clearance, but the corner distance is actually above the
clearance:
+---..
|
+---..
+---+
| |
: :
---
----
<Chris> Pad DRC has handling for trapezoid pads and handling for square
pads. This patch looks relatively low-risk to me, as square is a subset
of trapezoid and the handling for trapezoid pads is not specific to that
pad type.
* Remove download and build Boost from source and all CMake boost source build
dependencies.
* Make FindCairo.cmake use pkg-config when it's available not just all
platforms except windows.
* Add version checking to FindCairo.cmake.
* Change find GML version to 0.9.5.1 which is the current version in Ubuntu
14.04 LTS.
* Update required Cairo version to 1.12.1.
1. Fixed an assumption somebody originally made in the plugin that std::string had contiguous storage. This is not specced behavior pre C++11 so we gamble by calling .reserve() which should give a far better guarantee.
2. Added copy to clipboard information for curl
3. Removed some openssl references in compiling.md
4. Renamed struct vars to be uppercase to match "public var" code style policy
* fix minor coding style issues.
* remove unused member in modview_frame.
* use different frame names in ctor of modview_frame and viewlib_frame depending on the fact they are built modal or not modal (allowing identifications when calling wxFindWindowByName)
* change questionnable code in class_libentry.cpp.
* remove unused code in class_treeproject_item.cpp
Including the inline of the Output directory and its text field/browse button.
Moving the plot format to the left of the plot options as the plot options depend on the plot format.
In polygon calculations (combining polygons, fracture) the mode of calculation (fast or strictly simple polygon option) as no more a default value, because choosing the best mode is better to optimize the calculation time.
Maciej Suminski
John Beard
Cirilo Bernardo
jean-pierre charras
dino
Wayne Stambaugh
Simon Richter
Simon Wells
Dick Hollenbeck
decimad
Gustav Bergquist
Ian Roth
Chris Pavlina
Tomasz Wlostowski
Niki Guldbrand
Jon Neal
Chris Pavlina
Christian Gagneraud
Wayne Stambaugh
Thomas Pointhuber
Mrio Luzeiro
Alexander Lunev
Matthew Petroff
unknown
Patrick Pereira
Baranovskiy Konstantin
Cirilo Bernardo
unknown
Bernhard Stegmaier
unknown
Cirilo Bernardo
Rohan Agrawal
unknown
Tom Andrews
Andrew Zonenberg
Tomasz Wlostowski
Simon Wells
Element Green
unknown
unknown
Strontium