* Programmatically remove EnginePtr/inject, EXPAND_Engine, and WITH_ENGINE_TEST_HELPER*
for i in ": public EnginePtr " ", public EnginePtr" "EXPAND_Engine;" "EXPAND_Engine"; do
git grep -l "$i" | xargs sed -i "s/$i//g"
done
git grep -l "inject" | xargs sed -i "/inject[(][)]/d"
for i in WITH_ENGINE_TEST_HELPER_SENS WITH_ENGINE_TEST_HELPER_BOARD_CALLBACK WITH_ENGINE_TEST_HELPER; do
git grep -l "$i" | xargs sed -i "s/$i/EngineTestHelper eth/g"
done
git checkout firmware/controllers/core/engine_ptr.h
git checkout unit_tests/global.h
* Review fixups.
* Start the process of using a global engine ptr for unit tests.
Passing around the engine, config, and persistent state wasn't buying anything. Instead, use
EngineTestHelper(Base) to correctly set and clear the global variables. Add a dummy check in
case some test tries to set them manually.
* Fix OSX build
* Adapt PR to recently added code.
* static functions with (void)
* more
* Revert "more"
This reverts commit 246e53441f935451437df186ac92d7df26b62fb6.
* s
Co-authored-by: Matthew Kennedy <makenne@microsoft.com>
Better than reaching into members that should be private. I didn't feel like actually making
them private though, as one user validates pinStates isn't NULL.
Memory seems more valuable than CPU; use the O(lg n) lookup by angle. A side effect is that it
seems to fix trigger lookup for the second phase of TRIGGERTYPE 53 544 TT_TRI _TACH 0.00
Having to remove the precompiled header when switching from firmware to simulator to unit_tests and
back again is annoying. Use a directory to store the precompiled header output. Turn off
precompiled header warnings as GCC still complains even if it finds a valid one eventually.
The last version is already much better than the original, mostly because it doesn't call
CriticalSectionLocker, which has a ton of overhead due to debug tracking. But this version is
another 4 instructions / 12 bytes shorter. Does as much match in 32-bit land as possible, and
avoids math operations that span 64-bits (i.e. either operate on the lower half or upper half, but
not both). The result is only 3 instructions not including the necessary 4 loads (ptr to now, now,
ptr to upper, upper), 1 store (upper), and 1 return/branch.
Disallow scaled_channel for float. I can't think of a reason to allow it, and it gets in the way
of rounding.
Add separate template param to setTable; let the compiler sort out whether assignment can happen
between TElement and VElement without forcing them to be the same at function call time.
* scale map Y axis
* allow different row/col types
* scaled channel detector
* interpolation
* looks like this actually works
* tests, no manual scaling
* comment
* wiring
* static
* implementation
* comment
* unit tests happy
* guard
* guard smarter, not harder
* guard smarter not harder
* guard even smarter, not even harder
* don't need that
Scott Smith
rusefillc
Matthew Kennedy
Andrey