If the timer is not sufficiently high resolution but the graphics
card is fast, we can get renderTime.count() == 0.

Also, make sure it's localized properly.

After this commit, if the target system does have modern OpenGL
drivers installed, ANGLE is configured to use them, bypassing most
translation (shaders still have to be translated from ESSL to GLSL).

If there are no OpenGL drivers, such as if the graphics drivers were
installed via Windows Update, DirectX translation is still used. This
results in a very noticeable startup delay and minor performance
degradation.

In addition it is no longer necessary to build with -DOPENGL=1 to be
able to run the binary in wine; everything works out of the box.
Before, wine's incomplete HLSL translator would crash.

This change required renaming the variable `texture` in shaders,
since it shadows the Core GLSL function with the same name, and ANGLE
translates texture2D() calls to texture() calls.

This fixes an elusive GTK issue where tooltips would be spuriously
displayed, and makes tooltips behave nicer on Windows.

Unfortunately the macOS code is unchanged as the macOS tooltip
implementation seems seriously broken in ways I do not understand.

Before this commit it would prompt for top left and bottom left
corner, neither of which was what in fact was being used. Those two
specific points cannot be used because of the way equations are
written, so instead change that to top left and bottom right, which
is more convenient anyway.

This would add a lot of noise to the history, etc. if applied all
over now. Use git clang-format to apply it solely to your changes.

By setting WINVER=0x0501 (Windows XP) in CMakeLists.txt and adding a few
missing defines in guiwin.cpp and configuring OPENGL=1 in CMake
Solvespace (3.0~25b6eba1) compiles and works perfectly on Windows XP.
Tested with MinGW GCC-6.3.0-1

If a sketch has a "minor" problem, such as being self-intersecting,
this can cause considerably confusion in subsequent groups, yet is
not indicated in the group list.

This commit makes the "err" yellow in such cases. Note that the
indication may not change immediately when a change leading to
trouble is made, since the dependent groups are not recalculated
on all changes.

Also, mark not just curves, but also points and normals derived from
construction requests as construction.

Also, don't always mark arc center point as construction just to
exclude it from chord tolerance bounding box calculation; instead,
special-case it there.

This is currently necessary to get repeatable results when exporting
assemblies as a part of a batch process, since the mesh geometry in
imported files is not regenerated for export.

This is useful in niche cases, like making angular measurement tools.

Also, use simpler and more principled code for numeric precision
while editing constraints: don't special-case angles, but use up to
10 digits after the decimal point for everything.

Its only use was in a context where it was completely equivalent to
MemFree, so just use that instead, and keep the temporary heap as
purely an arena allocator, that could use something like bump
pointer.

Much clearer!

MSVC has a long history of value initialization bugs, and this one is
no exception. In this case, when some MSVC versions (at least up to
2013) are instructed to value-initialize a non-POD class with
a compiler generated non-trivial constructor, it does not zero out
the POD members.

It's not very obvious if the extrusion failed because in a later
group, the solid (by default) uses a very dark gray color that blends
into the black background.

This needs to be done separately because, while we already warn on
broken polygons in workplanes, many more groups can be extruded, e.g.
the canonical way (for now) to mirror a group is to use a rotation,
and that doesn't get checked for closed contour, since most rotations
won't get extruded.

This has always been possible by using negative scale, so this just
adds a checkbox controlling the sign.

Supported metric units: km, m, cm, mm, µm, nm.
Supported USCS units: in, mil, µin.

Also, use the newly introduced unit formatting machinery in tools for
measuring perimeter, area and volume, so that e.g. volume is not
displayed in millions of cubic millimeters.

Clarify the name of the command, as the old name is not strictly
correct. E.g. consider a vertical line with a midpoint constraint to
origin has 1 DOF, but 2 highlights are shown. Conversely, a single
datum point has 2 DOF, but 1 highlight is shown.

It makes no sense to solve by substitution (therefore weakening rank
check) in SolveRank(), since that's the whole point of SolveRank().

In addition, because SolveRank() is currently always called right
after AddConstraint(), forceDofCheck would always be true anyway.

In addition, it makes no sense to have TestRankForGroup() dependent
on the result of the previous solve. (For SolveGroup(), solving by
substitution after we know that rank test succeeds makes dragging
points much faster.)

This means that automatically added H/V constraints now will never
cause the sketch to become overconstrained, which currently makes
that feature almost unusable.

…oups.

This keeps groups with allowed redundant constraints cleaner when
they are used together with automatic constraints.

Otherwise it can result in a very confusing error that does not
suggest at all that the file is missing.

On a single load benchmark this provides about 25% speedup.

In other words, Ctrl inverts the normal action of LMB. It is already
possible to deselect entities through the context menu, but that
can be very awkward on laptop touchpads with a crowded sketch; with
Ctrl, a misclick is easily corrected without moving cursor at all.

Otherwise this results in harmless but annoying spam on applciation
shutdown.

See https://gitlab.gnome.org/GNOME/gtkmm/issues/22.

This was introduced in bd84bc1 and caused crashes with:
  Assertion failed: hm.v != t->h.v.