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
Currently, the TRY= comment needs to happen at the same time as the Run-TryBot+1 vote, otherwise it is silently ignored, which can be incredibly confusing (at least to me), and requires a not insignificant number of actions to resolve (deleting votes, re-commenting and voting, etc).
It would seem somewhat preferable to simply use the last TRY= comment on the change to determine slowbots (perhaps this requires retrieving more information from gerrit than the coordinator currently has?). Reseting the selected slowbows would require an empty TRY= if you wanted to disable them, but overall that seems less confusing (at least to me) to the current behavior.
Given the rapid approach of LUCI, this may not be particularly important, but it may help trigger something in the back on my mind next time I forget how to do this.
It would seem somewhat preferable to simply use the last TRY= comment on the change to determine slowbots (perhaps this requires retrieving more information from gerrit than the coordinator currently has?).
Yeah. The current suboptimal behavior is due to an implementation shortcut, see second sentence in the relevant comment. It was meant to be improved at some point so that "TRY= comments are ignored if they're not on the same comment that started the TryBots" could be deleted from https://go.dev/wiki/SlowBots#pitfalls.
At this point, as you said, LUCI's implementation of slowbots via a UI will likely supersede the TRY= comments and resolve this particular pitfall.
Currently, the
TRY=
comment needs to happen at the same time as the Run-TryBot+1 vote, otherwise it is silently ignored, which can be incredibly confusing (at least to me), and requires a not insignificant number of actions to resolve (deleting votes, re-commenting and voting, etc).It would seem somewhat preferable to simply use the last
TRY=
comment on the change to determine slowbots (perhaps this requires retrieving more information from gerrit than the coordinator currently has?). Reseting the selected slowbows would require an emptyTRY=
if you wanted to disable them, but overall that seems less confusing (at least to me) to the current behavior.Given the rapid approach of LUCI, this may not be particularly important, but it may help trigger something in the back on my mind next time I forget how to do this.
cc @golang/release @bcmills
The text was updated successfully, but these errors were encountered: