How not to submit a patchset?

What makes an apparently good patchset fail? Despite the fact it hasn’t been applied, did you know that working on this patchset might have actually been a great benefit for you and the kernel? A necessary step even?

Sometimes we involve into an idea that looks good at a first glance and might even look good to maintainers but eventually either:

  • Testing shows it’s not that awesome.
  • The benefit/maintainance-burden ratio is unclear.

So the patchset gets dropped in the end, or put aside waiting for better days. Then comes the time to realize how much you learned from it and how much the rest of the community also grew up from that failure.

Frederic Weisbecker

Frederic works for Suse. He has spent time on several parts of the kernel and is currently mostly focused on timers and RCU, with a growing taste for failing his own patchsets.