This reverts commit 68b1abf77f.
The warnings are valuable and shouldn't be cast aside.
As of 8f509f1, we special case macOS and don't set -fno-sanitize-recover
to allow CI to succeed.
In the future, this could be made stricter again by only suppressing
known bugs, which ideally should also be fixed or reported upstream.
Previously sanitizer flags were set unconditionally for
all code, including that of external libraries.
Set them only for targets in src/, tests/ and exposed/.
Unfortunately, the linker equivalent to add_compile_options,
add_link_options, is only available for CMake version >= 3.13.
So add the sanitizer flags manually to each target's linker options.
The only thing we need it anymore is the package version and platform
configuration, so only include it for that. As a result, less files
are rebuilt when the git commit changes and config.h is regenerated.
This is good practice and helps to catch bugs. Several changes
were made to accomodate the newly enabled warnings:
* -Wunused-function:
* in exposed/, static functions that were supposed to be inlined
were explicitly marked as inline;
* some actually unused functions were removed;
* -Wsign-compare: explicit conversions were added, and in
the future we should find a nicer way than aux* fields;
* -Wmissing-field-initializers: added initializers;
* -Wreorder: reordered properly;
* -Wunused-but-set-variable: remove variable.
-Wunused-parameter was turned off as enabling it would result in
massive amount of churn in UI code. Despite that, we should enable
it at some point as it has a fairly high SNR otherwise.
This will allow us to use non-POD classes inside these objects
in future and is otherwise functionally equivalent, as well
as more concise.
Note that there are some subtleties with handling of
brace-initialization. Specifically:
On aggregates (e.g. simple C-style structures) using an empty
brace-initializer zero-initializes the aggregate, i.e. it makes
all members zero.
On non-aggregates an empty brace-initializer calls the default
constructor. And if the constructor doesn't explicitly initialize
the members (which the auto-generated constructor doesn't) then
the members will be constructed but otherwise uninitialized.
So, what is an aggregate class? To quote the C++ standard
(C++03 8.5.1 §1):
An aggregate is an array or a class (clause 9) with no
user-declared constructors (12.1), no private or protected
non-static data members (clause 11), no base classes (clause 10),
and no virtual functions (10.3).
In SolveSpace, we only have to handle the case of base classes;
Constraint and Entity have those. Thus, they had to gain a default
constructor that does nothing but initializes the members to zero.
The GUI application and library share some object files but don't
enable object renaming resulting in the second compilation overwriting
the results of the first compilation. This can be fixed by setting
explicit preprocessor flags for one compilation. The actual problem,
however, is that recursive make is used in an inappropriate way; two
Makefiles shouldn't refer to the same source files. In this case,
libslvs should be under src/ and handled by the same Makefile, in
which case the issue would have been obvious when the Makefile was
written in the first place.
The libslvs library and CDemo program can now be built by Autotools. A few
code changes were needed for this: C++ comments in C code had to be
converted, constraint.cpp required some massaging, and fltkutil.cpp needed
a stub for InitHeaps().
This change comprehensively replaces the use of Microsoft-standard integer
and boolean types with their C99/C++ standard equivalents, as the latter is
more appropriate for a cross-platform application. With matter-of-course
exceptions in the Win32-specific code, the types/values have been converted
as follows:
QWORD --> uint64_t
SQWORD --> int64_t
DWORD --> uint32_t
SDWORD --> int32_t
WORD --> uint16_t
SWORD --> int16_t
BYTE --> uint8_t
BOOL --> bool
TRUE --> true
FALSE --> false
The following related changes are also included:
* Added C99 integer type definitions for Windows, as stdint.h is not
available prior to Visual Studio 2010
* Changed types of some variables in the SolveSpace class from 'int' to
'bool', as they actually represent boolean settings
* Implemented new Cnf{Freeze,Thaw}Bool() functions to support boolean
variables in the Registry
* Cnf{Freeze,Thaw}DWORD() are now Cnf{Freeze,Thaw}Int()
* TtfFont::Get{WORD,DWORD}() are now TtfFont::Get{USHORT,ULONG}() (names
inspired by the OpenType spec)
* RGB colors are packed into an integer of type uint32_t (nee DWORD), but
in a few places, these were represented by an int; these have been
corrected to uint32_t
the documentation accordingly. Also rename the C example for
consistency, and update copyright dates.
[git-p4: depot-paths = "//depot/solvespace/": change = 2190]
library. If I understand correctly, that will avoid all the
compiler version issues with different required versions of libc.
[git-p4: depot-paths = "//depot/solvespace/": change = 2167]
and cubics. Also add that to the library interface.
It might have been better to use a single constraint for that,
plus all the line-curve or line-line cases, but it would break
backwards compatibility if I did that now, and perhaps be
confusing with the 'other' member (which is meaningless for
lines) anyways.
[git-p4: depot-paths = "//depot/solvespace/": change = 2141]
one point to be dragged simultaneously. So now a dragged point
drags all the selected points and entities, and a dragged entity
drags its points (except for circles, which drag the radius).
This means that the number of forced points for the solver must now
be unlimited, and it is.
Also add commands to invert the selection within the active group,
and to select an edge chain starting from the current selection.
And redo the context menus a bit; still not great, but less
cluttered and more systematic.
[git-p4: depot-paths = "//depot/solvespace/": change = 2064]
surface's domain of u, v in [0, 1]. Cache the starting guess when
projecting a point into a ratpoly surface, to avoid brute force
searching for a good one every time. Split edges even if they
aren't quite inside the trim curve, since the trim boundaries are
pwl, not exact; unnecessary splits won't hurt, but failure to split
when necessary will. Make the triangulation code use a better (but
not perfect) epsilon, to avoid "can't find ear" failures on very
fine meshes.
And turn on compiler optimization! I had somehow forgotten about
that, and it's a ~2x improvement.
[git-p4: depot-paths = "//depot/solvespace/": change = 2026]
wrong parameters for a normal in Example2d(), and the horiz/vert
constraints were not documented correctly.
[git-p4: depot-paths = "//depot/solvespace/": change = 1963]
we're inconsistent (singular Jacobian). That's slow, so we should
provide a library interface to disable it.
[git-p4: depot-paths = "//depot/solvespace/": change = 1946]
little test app that links against it. I still need to polish a few
things, but this is more or less as it should be.
[git-p4: depot-paths = "//depot/solvespace/": change = 1944]
Maximilian Federle
Henrik Enggaard
luzpaz
whitequark
EvilSpirit
ruevs
Roland Lutz
Andrew Downing
Ilya Novoselov
Jonathan Westhues
Daniel Richard G