fix(engine): scan the Aura's own enchant-zone when choosing a host on non-cast entry#6072
Merged
matthewevans merged 1 commit intoJul 17, 2026
Conversation
… non-cast entry CR 303.4f: when an Aura enters the battlefield by any means other than resolving as a spell (e.g. returning from exile) and the effect putting it there doesn't specify a host, the controller chooses a legal object per the Aura's current Enchant ability. legal_aura_attachment_targets computed that candidate list from state.battlefield only, unconditionally - even though both call sites already had the Aura's enchant_filter in hand, and that filter is the single source of truth for which zone(s) it can legally target. An Enchant ability that targets "a creature card in a graveyard" (Animate Dead, Dance of the Dead), "an instant card in a graveyard" (Spellweaver Volute), or "a card in your hand" (Don't Worry About It) could therefore never find a legal host through this path, even when one genuinely existed - the consult always saw zero candidates and the Aura stayed stranded in its current zone (CR 303.4g). Rewrites legal_aura_attachment_targets to resolve the search zone(s) from enchant_filter.extract_zones() (defaulting to the battlefield when the filter imposes no zone constraint - the common case, unaffected), flat-map zone_object_ids over those zones, and filter with matches_target_filter - mirroring the existing object_count_matching_ids pattern in game/quantity.rs. Uses matches_target_filter rather than the spell-targeting enumerator: CR 115.1b says an Aura permanent doesn't target anything, so hexproof/shroud must never remove a candidate here, only an actual Enchant restriction can. Preserves the pre-existing "Enchant player" branch, the CR 303.4d self-enchant exclusion, and the can_attach_to_object legality gate unchanged. Incidentally also now excludes phased-out permanents from the default battlefield case (CR 702.26b) via zone_object_ids' existing filter - strictly more correct, not a behavior regression. This is the fix that closes the final leg of the Worldgorger Dragon self-loop combo (mass-return mechanism fixed separately in phase-rs#6055): the returning Animate Dead/Necromancy Aura can now find WGD sitting in the graveyard and re-attach to it, rather than staying stranded in exile.
Contributor
|
Warning You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again! |
rykerwilliams
added a commit
to rykerwilliams/phase
that referenced
this pull request
Jul 17, 2026
Parse changes introduced by this PR✓ No card-parse changes detected. |
matthewevans
approved these changes
Jul 17, 2026
matthewevans
left a comment
matthewevans
left a comment
Member
There was a problem hiding this comment.
Maintainer review: correct zone-authority fix with discriminating exile-return coverage; required checks are green.
rykerwilliams
added a commit
to rykerwilliams/phase
that referenced
this pull request
Jul 17, 2026
…reconciling WGD test next
rykerwilliams
added a commit
to rykerwilliams/phase
that referenced
this pull request
Jul 17, 2026
rykerwilliams
added a commit
to rykerwilliams/phase
that referenced
this pull request
Jul 17, 2026
merge origin/main (via a regular merge, not rebase, to preserve this PR's already-reviewed history) picks up phase-rs#6072, which fixed the returning Animate Dead Aura's graveyard-host search - the "known gap" this test's SCOPE section documented is now closed. Replaces the "stays in Exile" assertion with the actual now-correct outcome: the Aura reaches Zone::Battlefield and attaches to WGD-in-graveyard (CR 303.4f). Drives the rest of the pipeline to completion - the Aura's re-entry ETB and Piranha Marsh's ETB co-trigger (CR 603.3b OrderTriggers), Piranha's target selection and life-loss resolve first (CR 119.3, exact life-total delta), then the Aura's ETB resolves and reanimates Worldgorger Dragon a second time, re-attaching the Aura - closing one full cycle of the "Immortal Worldgorger Dragon" combo end-to-end. Also corrects a pre-existing CR citation: CR 610.3a ("the object doesn't move" edge case) was cited for the plain-land/Piranha-Marsh "return to the battlefield" behavior, which is the general CR 610.3 return rule, not the 610.3a edge case. Fixed in all 4 places this test cited it.
7 tasks
matthewevans
added a commit
to andriypolanski/phase
that referenced
this pull request
Jul 17, 2026
…ability pass (phase-rs#6055) * fix(engine): recognize mass-exile in the ETB-exile/LTB-return linked-ability pass CR 607.1 + CR 607.2a + CR 406.6: the engine already models "exile X, return X when the source leaves" generally (Journey to Nowhere / Oblivion Ring class) via a CR 610.3 "until"-duration vehicle - detect_etb_exile_ltb_return stamps Duration::UntilHostLeavesPlay onto the ETB exile at parse time, which makes zone_pipeline create a persistent ExileLinkKind::UntilSourceLeaves link per exiled object, and check_exile_returns (run every priority pass) performs the automatic return when the source leaves the battlefield - independent of whether the card's own printed LTB trigger ever resolves. trigger_is_etb_exile_pending_duration only matched the single-target Effect::ChangeZone->Exile, never the mass Effect::ChangeZoneAll->Exile that Worldgorger Dragon's "exile all other permanents you control" (and Realm Razer's "exile all lands") parses to. Their ETBs never got the duration stamp, so the whole downstream machinery stayed dormant and their LTB "return the exiled cards" triggers returned nothing. Widens the matches! to accept both shapes. apply_etb_exile_ltb_return needs no change - it only stamps execute.duration, never inspects execute.effect. Blast radius is bounded: detect_etb_exile_ltb_return only pairs an ETB exile with a card that also has a matching LTB-return trigger, so the widened arm cannot fire on a card that just permanently exiles things. Fixes the mass-return of non-Aura permanents in the Worldgorger Dragon + Animate Dead self-loop combo (a land returns untapped; Piranha Marsh's "target player loses 1 life" ETB refires) and fully fixes Realm Razer. The final leg that would close the infinite loop - the returning Animate Dead Aura re-attaching to Worldgorger-Dragon-in-graveyard - is NOT fixed here: the as-enters aura-host scan (legal_aura_attachment_targets) is battlefield-only and can't find a graveyard-resident host (CR 303.4f/g), so the Aura's re-entry is denied. That graveyard-host aura-attach gap is a separate, wider follow-up, tracked on the fork's internal worklist. * fix(engine): don't synthesize an exile-link return when the LTB rider carries an entry modifier CR 610.3: the automatic ExileLink::UntilSourceLeaves return path performs a plain, unmodified zone move - it can't carry a printed entry rider like "return the exiled cards to the battlefield tapped." Realm Razer's LTB return has exactly that rider (enter_tapped: true), so pairing it with the mass-exile synthesis from the previous commit would silently return its lands untapped, contradicting its printed text. trigger_is_ltb_return now also requires the return effect's entry modifiers (enter_tapped, enters_attacking, enter_transformed, owner_library, enters_under, enter_with_counters, conditional_enter_with_counters, face_down_profile, enters_modified_if) to all be at their no-op default. Realm Razer's ETB is therefore no longer paired and is left unstamped - correctly excluded rather than silently dropping its tapped modifier - until a modifier-carrying return mechanism exists. Worldgorger Dragon's LTB has no modifiers and is unaffected; it remains fixed. Adds a negative parser test asserting Realm Razer's mass-exile ETB is NOT stamped UntilHostLeavesPlay, and narrows the existing positive test to Worldgorger Dragon only. Updates the doc comments this touches (relation.rs, oracle.rs, casting_tests.rs) to drop the now-inaccurate "Realm Razer is fixed" claims and adds the CR 610.3 annotation prefix requested in review. Addresses maintainer review feedback on phase-rs#6055 (matthewevans): the mass-exile widening changed Realm Razer's required return semantics without carrying its tapped rider through the automatic return path. * test(engine): route the WGD test's off-zone keyword check through the state-scoped authority CI's engine-authorities gate flagged a raw obj.keywords.iter() query on an Exile-zone object - the object-scoped read is only authoritative for battlefield objects and silently misses off-zone keyword grants for hand/ graveyard/exile. Adds an explicit object_has_effective_keyword_kind(state, aura_id, KeywordKind::Enchant) check (the off-zone-aware authority) confirming the exiled Aura still carries an Enchant keyword at all, ahead of the existing raw structural inspection of which Enchant variant it is (its inner TargetFilter) - something no public authority exposes as a value, so that one line keeps the // allow-raw-authority annotation the gate's own escape hatch documents. * test(engine): close the Worldgorger Dragon self-loop test's final leg merge origin/main (via a regular merge, not rebase, to preserve this PR's already-reviewed history) picks up phase-rs#6072, which fixed the returning Animate Dead Aura's graveyard-host search - the "known gap" this test's SCOPE section documented is now closed. Replaces the "stays in Exile" assertion with the actual now-correct outcome: the Aura reaches Zone::Battlefield and attaches to WGD-in-graveyard (CR 303.4f). Drives the rest of the pipeline to completion - the Aura's re-entry ETB and Piranha Marsh's ETB co-trigger (CR 603.3b OrderTriggers), Piranha's target selection and life-loss resolve first (CR 119.3, exact life-total delta), then the Aura's ETB resolves and reanimates Worldgorger Dragon a second time, re-attaching the Aura - closing one full cycle of the "Immortal Worldgorger Dragon" combo end-to-end. Also corrects a pre-existing CR citation: CR 610.3a ("the object doesn't move" edge case) was cited for the plain-land/Piranha-Marsh "return to the battlefield" behavior, which is the general CR 610.3 return rule, not the 610.3a edge case. Fixed in all 4 places this test cited it. * fix(engine): mark a modifier-bearing linked return unsupported, not falsely green CR 610.3: for the ETB-exile/LTB-return linked-ability class, the actual return mechanism is the generic ExileLink::UntilSourceLeaves/ check_exile_returns path, driven purely by Duration::UntilHostLeavesPlay on the ETB - it performs an unmodified zone move and never reads anything from the LTB trigger's own printed effect. When the LTB return carries an entry rider the automatic path can't apply (Realm Razer's "return the exiled cards to the battlefield tapped"), the prior commit correctly declined to stamp the duration - but left no trace that this card is now unsupported, so it still parsed as a fully-formed, apparently-supported ETB/LTB pair. Splits trigger_is_ltb_return into a shape-only check plus the existing modifier check (trigger_is_ltb_return itself keeps its exact prior combined behavior - zero change for Journey to Nowhere / Oblivion Ring / Worldgorger Dragon), and adds trigger_is_ltb_return_with_entry_modifier for the shape-matched-but-modifier-rejected class. Parameterizes DocumentRelationIr::EtbExileLtbReturn with an outcome (DurationStamped, or ModifierUnsupported carrying the return's verbatim fragment text - captured at detect time since the applier has no access to the item list) rather than adding a sibling relation variant, per this module's own documented convention. When the outcome is ModifierUnsupported, attaches Effect::unimplemented("modifier_bearing_linked_return", fragment) to the LTB trigger's own execute chain, so coverage correctly reports the gap instead of the card showing as falsely supported. The new path is reachable only for a return the parser genuinely captured an entry modifier on, so it cannot false-flag a card that's actually fully supported - it only marks a class that was already broken (no duration stamp, no return) as visibly, honestly broken. Addresses maintainer review feedback on phase-rs#6055 (matthewevans, repeated across 5 rounds): mark this modifier-bearing linked-return combination unsupported through Effect::unimplemented and add a coverage-level regression test proving the gap is visible to coverage, not just that a duration field is None. --------- Co-authored-by: Matt Evans <1388610+matthewevans@users.noreply.github.com>
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
2 participants
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.
Summary
CR 303.4f: when an Aura enters the battlefield by any means other than resolving as a spell (e.g. returning from exile) and the effect putting it there doesn't specify a host, the controller chooses a legal object per the Aura's current Enchant ability.
legal_aura_attachment_targetscomputed that candidate list fromstate.battlefieldonly, unconditionally — even though both call sites already had the Aura'senchant_filterin hand, and that filter is the single source of truth for which zone(s) it can legally target.An Enchant ability that targets "a creature card in a graveyard" (Animate Dead, Dance of the Dead), "an instant card in a graveyard" (Spellweaver Volute), or "a card in your hand" (Don't Worry About It) could therefore never find a legal host through this path, even when one genuinely existed — the consult always saw zero candidates and the Aura stayed stranded in its current zone (CR 303.4g).
Fix: rewrite
legal_aura_attachment_targetsto resolve the search zone(s) fromenchant_filter.extract_zones()(defaulting to the battlefield when the filter imposes no zone constraint — the common case, unaffected), flat-mapzone_object_idsover those zones, and filter withmatches_target_filter— mirroring the existingobject_count_matching_idspattern ingame/quantity.rs. Usesmatches_target_filterrather than the spell-targeting enumerator: CR 115.1b says an Aura permanent doesn't target anything, so hexproof/shroud must never remove a candidate here, only an actual Enchant restriction can. Preserves the pre-existing "Enchant player" branch, the CR 303.4d self-enchant exclusion, and thecan_attach_to_objectlegality gate unchanged.Incidentally also now excludes phased-out permanents from the default battlefield case (CR 702.26b) via
zone_object_ids's existing filter — strictly more correct, not a behavior regression.This is the fix that closes the final leg of the Worldgorger Dragon self-loop combo (mass-return mechanism fixed separately in #6055): the returning Animate Dead/Necromancy Aura can now find WGD sitting in the graveyard and re-attach to it, rather than staying stranded in exile.
Test plan
exile_return_tests(16/16),animate_dead-filtered casting tests (9/9),necromancy-filtered casting tests (5/5)cargo fmt --all,cargo clippy -p engine --all-targets -- -D warningsclean