Repository navigation
Conversation
The leg's `to` was the walk-29 quiet point north-east of the temple and its second via walked the party back to it before the approach path turned south: a run starting at the leg's first row doubled back (leg 7 end -> pull 239.3 yd; the JSON chain placement -> via -> to -> boss approach -> boss 371.1 yd). `to` is the row a placed Hakkar run already starts on (-11786.999 -1564.2667 20.562721) and the first via keeps the connection from leg 7's end: leg 7 end -> pull 149.8 yd, chain 290.1 yd. The pull groups and the boss step are unchanged.
…ck's second round Walk 71 paid 115 `q inst` reads and 114 `.go xyz` + 2 s sleeps for 52 packs: every pack was walked twice, and the second round read the same dead units (the four one-member packs of a formation group each got their own pair of visits). The instance is now read once per pack (the end of the clear still reads it once more), and the second round - and the rest of a round's stops - is skipped once every member is dead or has had its four kill lines. A member never seen or seen alive still gets the second round, and a member is skipped exactly as before, so the killed set is unchanged. ~120 s of the 322 s gear-to-place median goes. Tests: TestPlaceGMClearSkipsTheSecondRoundOfADeadPack (fails at 2 clear reads before the change; disabling the skip reproduces it), and the wide-pack test now pins a wholly empty first round still reading round 2 and exactly 2 `q inst` in the clear. docs/okf/raids/place-mid-raid.md says the new rule.
The once-per-pack read of 9a9ac8f skipped placeInstanceIs after a pack's first stop, so a later stop landing in another copy still read units and sent irreversible GM kills there. The check is back after every .go xyz and before every units read or kill: a mismatch stops the clear on the existing error path with no kill sent. Only the dead-pack second-round skip of 9a9ac8f stays: a pack whose every member is dead (or used its four kill lines) after the first round is not walked again. Walk 71's ~230 s of the 322 s pre-clear for those empty second rounds still goes; the per-stop instance reads are back. Tests: TestPlaceGMClearSecondStopInstanceMismatchStopsTheClear (a second stop in another copy: no second-stop read or kill, the clear stops; drop the per-stop check and it fails at the post-move read), kept TestPlaceGMClearSkipsTheSecondRoundOfADeadPack, and the wide-pack test now pins one instance read per stop plus the final check. docs/okf/raids/place-mid-raid.md says the restored rule.
origins_test.go: the walk-71-fast pin comment counts export rows the way the older entries do (export 694 -> 691 rows), and names the moved to (old -11754.643 -1494.6427 37.765 -> new -11786.999 -1564.2667 20.563) and the dropped via (-11776.531 -1502.4786 29.246). Comment only: the pin value 62b90de7 stays, and the TortoiseBots export (migration 20261111100000, --rev 9a9ac8f, 691 steps) still matches it. dungeons/zul_gurub.json: the walk-71 note gains the plan-check line for the new via->to hop (plan-check --dir dungeons --mmaps ~/turtle-agent-mmaps zul_gurub with a VM-built binary: ok, 851 points, 0 issues; export steps 400 -> 401, 55.2 yd, no finding) and the patrol-loop margin (the new to is 55.5 yd from the nearest patrol loop, 11357 Son of Hakkar guid 49033). Notes are not exported: the pin still passes unchanged. Rows untouched. Tests: TestPlaceGMClearSecondRoundKillsAMemberMissedInRoundOne (round 1 misses the far member, round 2 kills it; removing the second round fails at partial true) and TestPlaceGMClearFinalInstanceCheckFailureSendsNoKill (the final check names another copy: exit 18, no kill after; dropping the check fails at exit 0, want 18).
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Faster Hakkar attempts in Zul'Gurub raid walks: a quicker GM pre-clear and a Hakkar leg that no longer doubles back.
What changes
raid-walk --gm-clear(cmd/turtle-agent/raid_walk_gmclear.go): a pack whose first round found every member dead (or given up) skips its second round. The instance is still read after every.go xyzand before any unit read or kill. A stop that lands in another instance copy stops the clear with no kill.dungeons/zul_gurub.json, leghakkar: the leg'stois the placement row the placed walk already starts on (-11786.999 -1564.2667 20.563), and the second via is dropped. Leg 7's end → first pull goes from 239.3 yd to 149.8 yd; the export goes from 694 to 691 rows. The origins pin is updated. The matching TortoiseBots migration is20261111100000_world.sql(--rev 9a9ac8f0); its rows match this pin.Why
In walk 71, an attempt spent a median 322 s from the gear line to placement, 260 s of it in the clear: every pack was visited twice, with a 2 s pause per stop. A raid entering the Hakkar leg also walked north-east and back. The measurements are in the walk 71 timeline analysis.
Tests
vm-test ta: PASS on c7abe68;gofmtis clean.Reviews
Status
Draft until walk 72 shows the faster setup live (gear → place seconds, place → pull seconds, Hakkar fights per hour). The walk binary is
~/ta-1111a-bin(c7abe68).