Pj/task collision test #79
Conversation
| - name: test platforms | ||
| #run: python3 ci/build_platform.py main_platforms | ||
| run: python3 ci/build_platform.py main_platforms esp32 esp8266 | ||
| run: python3 ci/build_platform.py main_platforms |
There was a problem hiding this comment.
main_platforms now support/check esp32 and esp8266.
| { | ||
| struct task * const t = static_cast<struct task * const>(task); | ||
|
|
||
| if (t) { |
There was a problem hiding this comment.
Sanitize inputs: t is never 0 now, and check t is in tasks[] bounds
|
I was concerned that this test would take a lot of time. It is not as bad as I thought.
|
| Timer<0x2, CLOCK::millis, Task *> timer; | ||
|
|
||
| assertEqual(timer.size(), 0UL); | ||
| assertEqual((unsigned long) timer.size(), 0UL); |
There was a problem hiding this comment.
I run my test on an Arduino Nano. I got an error because size_t is NOT unsigned long. This cast should make the test work on all platforms.
|
Looks good - can we consolidate the commits into fewer, more concise chunks? |
I will see what I can do. Part of the reason for the smaller commits is that I usually write code using a Windows computer, so it is more difficult to test how the EpoxyDuino platform will behave without trying it out with the GitHub Action -- I had to use a commit to test remotely. BUT I can squash these changes to another branch before setting up a pull request for you to review. |
see if redundant task IDs allocated see if NULL task ID created see if a double cancel() kills another task makefile rule for timerMultiCancelTest EpoxyDuino doesn't support serial print HEX.
cancel() returns success only if it removed a task check cancel with stale pointer fails cancelling a removed task should fail add range check for cancel() arg
philj404
force-pushed
the
pj/taskCollisionTest
branch
from
63c4d30 to
a4c075b
Compare
|
I consolidated the PR into 4 commits total (rebased to 3 commits, plus a new commit for your parenthesis suggestion). This is the first time I have used |
2 participants
I decided to create a pull request for my regression test of timer ID collisions, cancel() etc.
... and, I think I discovered a regression.
Now that
timer.cancel(anId)returns abool, AND timers no longer use a counter to randomize Task values, timer.cancel() will always find a Task to cancel, even if it is an empty task.So...
timer.cancel(anId)always returnstrue... even if anId is cancelled twice in a row.