Fix inaccurate timeouts

[aentwist]

Timeouts are currently measured by,

for timeout / refresh_rate times do
    is_done <- body()
    if is_done then break
    wait refresh_rate
end for

However, the execution time of the body is not always negligible. Especially for small refresh rates the body time becomes extremely significant, in some cases ~6-60x the refresh rate. In these cases the wall time of the timeout is proportionally affected. That is, timeouts become 6-60x longer than specified.

Instead timeouts should be measured using a pure wall time approach,

t_start <- get_time()
t_end <- get_time()
while t_end - t_start < timeout do
    is_done <- body()
    if is_done then break
    wait refresh_rate

    t_end <- get_time()
end while

By factoring in the body time, this should keep them accurate.

[aentwist]

The suggested fix has an inaccurate refresh rate

[aentwist]

🎉 This issue has been resolved in version 1.1.0 🎉

The release is available on:

• v1.1.0
• GitHub release

Your semantic-release bot 📦🚀