You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
As a counter-proposal to #17503, we could improve STW times by performing stack re-scanning concurrently. I'm not actually proposing this, but I wrote a whole design doc before realizing a simpler approach would let us completely eliminate stack re-scanning. This issue is a place to put that design doc.
The text was updated successfully, but these errors were encountered:
Thanks for playing this out, @aclements. I agree that the approach in #17503 is much simpler—to understand, to reason about, to implement. The transitive mark write barrier seems like it would invite tail latency bugs, which would be hard to diagnose and to work around.
@rhysh, thanks for the suggestion in the first place. :) Writing out this design in depth is what led me to #17503 (which is not as simple as it sounds to implement, but is very nearly done anyway).
gopherbot
pushed a commit
to golang/proposal
that referenced
this issue
Oct 24, 2016
As a counter-proposal to #17503, we could improve STW times by performing stack re-scanning concurrently. I'm not actually proposing this, but I wrote a whole design doc before realizing a simpler approach would let us completely eliminate stack re-scanning. This issue is a place to put that design doc.
The text was updated successfully, but these errors were encountered: