go-pkgz / ctrl / 32208359791
95%
master: 95%

LAST BUILD BRANCH: refs/tags/v0.2.1
DEFAULT BRANCH: master
Ran
Jobs 1
Files 4
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 94.924%. Remained the same
32208359791

push

github

web-flow 
Replace timing-based waits in tests with synchronisation (#5)

Previously, the shutdown tests slept for a fixed period and then asserted that
the injected exit function had been called, which fails on a loaded machine
whenever the scheduler runs the timer later than the sleep. The injected exit
now reports through a channel, so each test waits exactly as long as it needs
to and fails on a deadline instead.

The second-signal case ran with a 500ms shutdown timeout while asserting on an
exit that the second signal was meant to cause, so the timeout could produce
that exit on its own; its timeout is now long enough that only the signal can.
The concurrent-signals case keeps a short timeout on purpose, as os/signal
drops a send to a full channel and a burst may therefore be delivered once, but
it now runs on a signal of its own and joins its senders, so nothing it sends
can reach a later case.

The timeout accuracy case measured from before the signal that starts the timer
and asserted a 150ms window around it. It now measures from the shutdown
callback, runs on its own signal, keeps the meaningful bound that the forced
exit never happens early, and leaves the upper one loose enough to survive a
delayed timer.

The HTTP tests no longer sleep before the first request, as the listener is
created before the server starts and already accepts connections; their
requests carry a client timeout, so a stuck server fails the test instead of
hanging it.

187 of 197 relevant lines covered (94.92%)

5.61 hits per line

Jobs
ID Job ID Ran Files Coverage
1 32208359791.1 4
94.92
 GitHub Action Run
Source Files on build 32208359791