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
Filtering commits to a path was done in a inefficient way (on the fly check diffs for matching paths). The current version walks the tree once to gather all commit hashes so it can filter by those instead (much faster on-the-fly), however this introduces a significant startup cost (~6s on a 7752 commit tree).
There must be smarter ways to do this because gitk startup time for the same task is barely noticeable.
The text was updated successfully, but these errors were encountered:
Trying to fix#1, the high level idea here is to only walk the revision
graph once, as the graph is walked then commits are also stored in a
cache so that if the user goes back up the history we don't have to
start a new walker. Instead older commits are served from the cache.
This would mean we could probably remove the startup tree scan for
matching commits entirely at the cost of growing memory usage over time.
However, I painted myself into a corner /w the borrow checker and
probably need to revisit an abstraction here.
Filtering commits to a path was done in a inefficient way (on the fly check diffs for matching paths). The current version walks the tree once to gather all commit hashes so it can filter by those instead (much faster on-the-fly), however this introduces a significant startup cost (~6s on a 7752 commit tree).
There must be smarter ways to do this because
gitk
startup time for the same task is barely noticeable.The text was updated successfully, but these errors were encountered: