After releasing v1.2.1 to correctly handle slow_down errors from the OAuth server, we're now seeing endless loops of increasing wait times in gh.
Thanks to our community who tried the gh v2.85.0, we noticed in WSL environments (and possibly with other virtualisation technologies), the monotonic clock runs faster than real time, resulting in actually shorter polling intervals, and hence repeated slow_down errors. See cli/cli#9370 (comment) and discussion that follows for more context.
We should now improve the polling interval precision to accommodate for a faster ticking monotonic clock, if any. Note that if the monotonic clock runs slow, we wouldn't have any problem as we'd actually wait more than needed to poll the server again.
After releasing
v1.2.1to correctly handleslow_downerrors from the OAuth server, we're now seeing endless loops of increasing wait times ingh.Thanks to our community who tried the gh
v2.85.0, we noticed in WSL environments (and possibly with other virtualisation technologies), the monotonic clock runs faster than real time, resulting in actually shorter polling intervals, and hence repeatedslow_downerrors. See cli/cli#9370 (comment) and discussion that follows for more context.We should now improve the polling interval precision to accommodate for a faster ticking monotonic clock, if any. Note that if the monotonic clock runs slow, we wouldn't have any problem as we'd actually wait more than needed to poll the server again.