Tyler Laprade
04/15/2026, 9:06 PMPylon
04/15/2026, 9:06 PMgt feedback --with-debug-context <any message>Tyler Laprade
04/15/2026, 9:06 PMgt get 1234 --force used to get the parent tracking from remoteTyler Laprade
04/15/2026, 9:06 PM--force flagPylon
04/15/2026, 9:15 PMgt feedback --with-debug-context <message> so i can take a look at whats happening in our logs?
we did release 1.8.4 today which included an update to gt get . are you running on the newest version or are you on 1.8.3 still?Tyler Laprade
04/15/2026, 9:16 PMPylon
04/15/2026, 9:35 PMPylon
04/16/2026, 3:08 PM--force does re-pull parent tracking from remote but only on the code path gt get takes when the PR isn't already cached locally. Because #8382 is cached on your machine with a local branch that isn't Graphite-tracked, gt get takes a different path that hits the tracking check first and bails before --force gets applied. Running gt track --parent <parent> once will get you back on the happy path.
this also happens to our own engineering team and there's a ticket that's in progress right now on how we handle branches not tracked yet by graphite. i dont have an ETA yet for when the fix will be released.Tyler Laprade
04/16/2026, 3:09 PMgt track --parent <parent> is not an option since this happens in automated scripts and the parent isn’t necessarily known.Tyler Laprade
04/16/2026, 3:09 PMgt get with no PR number is always going to be cached locally, right? So that check doesn’t quite make sense.Tyler Laprade
04/16/2026, 3:09 PMgit instead of `gt`…Pylon
04/16/2026, 3:12 PMgit or gt. since i just started at graphite, my muscle memory has been doing git instead of gtTyler Laprade
04/16/2026, 3:36 PMgt for everything when possible, but some of my scripts do git for worktree management before handing off to gt, which is likely why I keep running into thisTyler Laprade
04/16/2026, 3:37 PM--force to have consistent behavior for gt get 1234 --force and gt get --force when already on the branch for PR 1234