profile
viewpoint

startedapple/swift-lldb

started time in 25 minutes

startedapple/swift-evolution

started time in 33 minutes

CommitCommentEvent

push eventapple/foundationdb

Evan Tschannen

commit sha 9ec9f41d34da401ba7d7a81bc87710004a401ac6

backup and DR would not share mutations if they were started on different versions of FDB

view details

Evan Tschannen

commit sha b88b1614c136bb27f6bbee87bffb93e1bd1215e4

bool knobs would improperly report a SevWarnAlways that they were unrecognized

view details

Evan Tschannen

commit sha 628b4e0220fe435f8fb5c62b4399d0a38460fb0a

added a warning if multiple log ranges exist for the same range

view details

Evan Tschannen

commit sha 8b09cd16b25df20b656e3eb7fdbe94801e294ce4

Merge branch 'release-6.2' of github.com:apple/foundationdb into feature-share-mutations

view details

Evan Tschannen

commit sha 2facfc090bf0fd59f79399a6ac3a0dfc5ac8f501

Update fdbserver/Status.actor.cpp Co-Authored-By: A.J. Beamon <ajbeamon@users.noreply.github.com>

view details

Evan Tschannen

commit sha 5be773f145687976cc1f66c3d838da28c1a0701f

Update fdbserver/Status.actor.cpp Co-Authored-By: A.J. Beamon <ajbeamon@users.noreply.github.com>

view details

Evan Tschannen

commit sha 587cbefe7fdeb8e1b2af3d918a843f36f60a9ffb

duplicate mutation stream checker did not have a timeout duplicate mutation stream did not work properly when multiple ranges exist with the same begin key

view details

Evan Tschannen

commit sha 1fd9411395becf5089688b3e027509bd2198168b

fix: eraseLogData did not cleanup destUidLookupPrefix

view details

Evan Tschannen

commit sha 42b7acf7b7e886beaba2e4cc4038838bbb17582c

Merge pull request #2202 from etschannen/feature-share-mutations Backup and DR would not share mutations if started on different versions of FDB

view details

push time in 34 minutes

pull request commentapple/foundationdb

Backup and DR would not share mutations if started on different versions of FDB

It is possible backup will need be started and aborted multiple times to get backup and DR to share, because the destUIDLookup could actually have ranges from more than 2 versions of FDB (although it would only be saving the mutations for two of those lookup entries). When backup is started it will randomly join with one of the lookups for the same range, and if that happens to be the active DR then it will just work. If it is not the DR one, when you abort the backup it will remove the lookup key associated with the backup, so the next time backup starts it will join with a different lookup.

etschannen

comment created time in 34 minutes

push eventapple/llvm-project-v4

Sam Clegg

commit sha 6c393e9d74d27a9e9d0d9c60537d7b2d4203ae13

[lld][WebAssembly] Fix for weak references to data symbols in archives Fix a bug where were not handling relocations against weakly undefined data symbol. Add a test for this case. Also ensure that the weak references to data symbols are not pulled in from archive files by default (but are if `-u <name>` is added to the command line). Fixes: PR43696 Differential Revision: https://reviews.llvm.org/D69073 llvm-svn: 375077

view details

push time in 38 minutes

push eventapple/llvm-project-v2

Sam Clegg

commit sha 6c393e9d74d27a9e9d0d9c60537d7b2d4203ae13

[lld][WebAssembly] Fix for weak references to data symbols in archives Fix a bug where were not handling relocations against weakly undefined data symbol. Add a test for this case. Also ensure that the weak references to data symbols are not pulled in from archive files by default (but are if `-u <name>` is added to the command line). Fixes: PR43696 Differential Revision: https://reviews.llvm.org/D69073 llvm-svn: 375077

view details

push time in 40 minutes

push eventapple/foundationdb

Evan Tschannen

commit sha 35ef9b32deabe5a76c5e00ccb4ae62dd10b10761

fix: if establishing a TLS connection took longer than 10ms, we could spend all our CPU establishing new connections instead of pinging to maintain existing connections, leading to an infinite loop

view details

Evan Tschannen

commit sha b23eaa1b8412944c02e1f6cd69c90b9992111f6c

Merge pull request #2252 from etschannen/feature-increase-accept-delay Increase connection_accept_delay

view details

push time in 40 minutes

push eventapple/foundationdb

Evan Tschannen

commit sha ac28e96bbf110b244fb50b0a6523c28ed84bc87b

added a yield on the proxy to remove a slow task when processing large transactions

view details

Evan Tschannen

commit sha a81ff63147b1262d361fc78b32aeeaa62d3c2a7b

Merge pull request #2250 from etschannen/feature-fix-proxy-slow-task added a yield on the proxy to remove a slow task when processing large transactions

view details

push time in 41 minutes

Pull request review commentapple/swift

[AutoDiff] Add `differentiability_witness_function` instruction.

 Extracts the original function or a transpose function from the given ``[original]`` or ``[transpose]``.  +differentiability_witness_function+``````````````````````````````````++::++  sil-instruction ::= 'differentiability_witness_function'+                      sil-differentiability-witness-function-kind
                      '[' sil-differentiability-witness-function-kind ']'
dan-zheng

comment created time in an hour

Pull request review commentapple/swift

[AutoDiff] Add `differentiability_witness_function` instruction.

 Extracts the original function or a transpose function from the given ``[original]`` or ``[transpose]``.  +differentiability_witness_function+``````````````````````````````````++::++  sil-instruction ::= 'differentiability_witness_function'+                      sil-differentiability-witness-function-kind+                      '[' 'parameters' sil-differentiability-witness-indices ']'+                      '[' 'results' sil-differentiability-witness-indices ']'+                      generic-parameter-clause?+                      sil-function-name ':' sil-type++  sil-differentiability-witness-function-kind ::=+      '[' sil-differentiability-witness-function-kind-name ']'+  sil-differentiability-witness-function-kind-name ::=+      'jvp' | 'vjp' | 'transpose'+  sil-differentiability-witness-indices ::= [0-9]+ (' ' [0-9]+)*
  sil-differentiability-witness-function-index-list ::= [0-9]+ (' ' [0-9]+)*
dan-zheng

comment created time in an hour

Pull request review commentapple/swift

[AutoDiff] Add `differentiability_witness_function` instruction.

 Extracts the original function or a transpose function from the given ``[original]`` or ``[transpose]``.  +differentiability_witness_function+``````````````````````````````````++::++  sil-instruction ::= 'differentiability_witness_function'+                      sil-differentiability-witness-function-kind+                      '[' 'parameters' sil-differentiability-witness-indices ']'+                      '[' 'results' sil-differentiability-witness-indices ']'+                      generic-parameter-clause?+                      sil-function-name ':' sil-type++  sil-differentiability-witness-function-kind ::=+      '[' sil-differentiability-witness-function-kind-name ']'+  sil-differentiability-witness-function-kind-name ::=+      'jvp' | 'vjp' | 'transpose'+  sil-differentiability-witness-indices ::= [0-9]+ (' ' [0-9]+)*+  generic-parameter-clause ::=

Add a "sil-differentiability-witness-function-" prefix to all non-terminals.

  sil-differentiability-witness-function-generic-parameter-clause ::=
dan-zheng

comment created time in an hour

Pull request review commentapple/swift

[AutoDiff] Add `differentiability_witness_function` instruction.

 Extracts the original function or a transpose function from the given ``[original]`` or ``[transpose]``.  +differentiability_witness_function+``````````````````````````````````++::++  sil-instruction ::= 'differentiability_witness_function'+                      sil-differentiability-witness-function-kind+                      '[' 'parameters' sil-differentiability-witness-indices ']'+                      '[' 'results' sil-differentiability-witness-indices ']'+                      generic-parameter-clause?+                      sil-function-name ':' sil-type++  sil-differentiability-witness-function-kind ::=

Remove this non-terminal.

dan-zheng

comment created time in an hour

Pull request review commentapple/swift

[AutoDiff] Add `differentiability_witness_function` instruction.

 Extracts the original function or a transpose function from the given ``[original]`` or ``[transpose]``.  +differentiability_witness_function+``````````````````````````````````+
dan-zheng

comment created time in an hour

push eventapple/llvm-project-v3

Sam Clegg

commit sha 6c393e9d74d27a9e9d0d9c60537d7b2d4203ae13

[lld][WebAssembly] Fix for weak references to data symbols in archives Fix a bug where were not handling relocations against weakly undefined data symbol. Add a test for this case. Also ensure that the weak references to data symbols are not pulled in from archive files by default (but are if `-u <name>` is added to the command line). Fixes: PR43696 Differential Revision: https://reviews.llvm.org/D69073 llvm-svn: 375077

view details

push time in 42 minutes

push eventapple/llvm-project-v2-split

Marcello Maggioni

commit sha 95a064c8ac3ade9fa555916c61b460836eaaae73

Move LiveRangeCalc header to publicily available position. NFC Differential Revision: https://reviews.llvm.org/D69078 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@375075 91177308-0d34-0410-b5e6-96231b3b80d8

view details

Marcello Maggioni

commit sha 8326c41046684e0d417bdc8eb0dbb768c06c866d

Clang-formatting of some files in LiveRangeCalc header (LiveRangeCalc.h) NFC git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@375076 91177308-0d34-0410-b5e6-96231b3b80d8

view details

push time in an hour

push eventapple/llvm-project-v4-split

Marcello Maggioni

commit sha 95a064c8ac3ade9fa555916c61b460836eaaae73

Move LiveRangeCalc header to publicily available position. NFC Differential Revision: https://reviews.llvm.org/D69078 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@375075 91177308-0d34-0410-b5e6-96231b3b80d8

view details

Marcello Maggioni

commit sha 8326c41046684e0d417bdc8eb0dbb768c06c866d

Clang-formatting of some files in LiveRangeCalc header (LiveRangeCalc.h) NFC git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@375076 91177308-0d34-0410-b5e6-96231b3b80d8

view details

push time in an hour

Pull request review commentapple/swift

[TBDGen] Skip non-inlinable function bodies in InstallAPI

  // 1. Emit two TBDs, one with -tbd-is-installapi set, and one without -// RUN: %target-swift-frontend -emit-ir -o /dev/null %s -tbd-is-installapi -emit-tbd -emit-tbd-path %t/flag-provided.tbd+// RUN: %target-swift-frontend -typecheck -o /dev/null %s -tbd-is-installapi -emit-tbd -emit-tbd-path %t/flag-provided.tbd // RUN: %target-swift-frontend -emit-ir -o /dev/null %s -emit-tbd -emit-tbd-path %t/flag-omitted.tbd

Yeah, I just meant the comment makes it sound like they're supposed to be the same except for this one flag.

harlanhaskins

comment created time in an hour

startedapple/swift

started time in an hour

push eventapple/llvm-project-v3

Marcello Maggioni

commit sha 6fc9563dba6319ab0715c6fd4cb072dc6edc4c72

Move LiveRangeCalc header to publicily available position. NFC Differential Revision: https://reviews.llvm.org/D69078 llvm-svn: 375075

view details

Marcello Maggioni

commit sha 5926eb5fe64ac9e2c1324811a0cc0d7bd5fb2da7

Clang-formatting of some files in LiveRangeCalc header (LiveRangeCalc.h) NFC llvm-svn: 375076

view details

push time in an hour

push eventapple/llvm-project-v4

Marcello Maggioni

commit sha 6fc9563dba6319ab0715c6fd4cb072dc6edc4c72

Move LiveRangeCalc header to publicily available position. NFC Differential Revision: https://reviews.llvm.org/D69078 llvm-svn: 375075

view details

Marcello Maggioni

commit sha 5926eb5fe64ac9e2c1324811a0cc0d7bd5fb2da7

Clang-formatting of some files in LiveRangeCalc header (LiveRangeCalc.h) NFC llvm-svn: 375076

view details

push time in an hour

push eventapple/llvm-project-v2

Marcello Maggioni

commit sha 6fc9563dba6319ab0715c6fd4cb072dc6edc4c72

Move LiveRangeCalc header to publicily available position. NFC Differential Revision: https://reviews.llvm.org/D69078 llvm-svn: 375075

view details

Marcello Maggioni

commit sha 5926eb5fe64ac9e2c1324811a0cc0d7bd5fb2da7

Clang-formatting of some files in LiveRangeCalc header (LiveRangeCalc.h) NFC llvm-svn: 375076

view details

push time in an hour

startedapple/swift-corelibs-foundation

started time in an hour

startedapple/swift

started time in an hour

pull request commentapple/swift

Further simplifications to omitNeedlessWords

@swift-ci Please test source compatibility

jrose-apple

comment created time in an hour

pull request commentapple/swift

Further simplifications to omitNeedlessWords

@swift-ci Please test

jrose-apple

comment created time in an hour

PR opened apple/swift

Further simplifications to omitNeedlessWords

Follow-up to #27602, mostly in other parts of the algorithm. No intended functionality change, although I messed that up last time.

+80 -101

0 comment

4 changed files

pr created time in an hour

Pull request review commentapple/foundationdb

Add fdbcli and API command to mark storage servers as permanently failed

 struct RemoveServersSafelyWorkload : TestWorkload {  		state std::vector<ISimulator::ProcessInfo*>	killProcArray; 		state std::vector<AddressExclusion>	toKillArray;-+		state std::vector<AddressExclusion>	toKillMarkFailedArray;+		state AddressExclusion coordExcl;+		// Exclude a coordinator under buggify, but only if fault tolerance is > 0 and kill set is non-empty already+		if (BUGGIFY && toKill.size()) {+			std::vector<NetworkAddress> coordinators = wait(getCoordinators(cx));+			if (coordinators.size() > 2) {+				auto randomCoordinator = deterministicRandom()->randomChoice(coordinators);+				coordExcl = AddressExclusion(randomCoordinator.ip, randomCoordinator.port);+			}+		} 		std::copy(toKill.begin(), toKill.end(), std::back_inserter(toKillArray));+		if (markExcludeAsFailed) {+			state int retries = 0;+			loop {+				state bool safe = false;+				state std::set<AddressExclusion> failSet =+				    random_subset(toKillArray, deterministicRandom()->randomInt(0, toKillArray.size() + 1));+				if (coordExcl.isValid()) {+					failSet.insert(coordExcl);+				}+				toKillMarkFailedArray.resize(failSet.size());+				std::copy(failSet.begin(), failSet.end(), toKillMarkFailedArray.begin());+				TraceEvent("RemoveAndKill", functionId)+				    .detail("Step", "SafetyCheck")+				    .detail("Exclusions", describe(toKillMarkFailedArray));+				choose {+					when(bool _safe = wait(checkSafeExclusions(cx, toKillMarkFailedArray))) {+						safe = _safe;+					}+					when(wait(delay(5.0))) {+						TraceEvent("RemoveAndKill", functionId)+							.detail("Step", "SafetyCheckTimedOut")+							.detail("Exclusions", describe(toKillMarkFailedArray));+					}+				}+				if (retries == self->maxSafetyCheckRetries) {+					// Do not mark as failed if limit is reached+					TraceEvent("RemoveAndKill", functionId)+						.detail("Step", "SafetyCheckLimitReached")+						.detail("Retries", retries);+					markExcludeAsFailed = false;+					safe = true;+				}+				if (safe) break;+				retries++;+			}+		}+		// Swap out coordinator with server in kill set, but only if already marking as failed and safety check passes

Doesn't the set of processes to kill that passes the safetyCheck already include the coordinator? See https://github.com/apple/foundationdb/blob/f4237ebffffad14f370fc1c99ac241bbe3c086d9/fdbserver/workloads/RemoveServersSafely.actor.cpp#L423, the coordExcl is added to failSet and the failSet is passed into checkSafeExclusions.

Did I miss something?

fzhjon

comment created time in an hour

Pull request review commentapple/swift

[TBDGen] Skip non-inlinable function bodies in InstallAPI

  // 1. Emit two TBDs, one with -tbd-is-installapi set, and one without -// RUN: %target-swift-frontend -emit-ir -o /dev/null %s -tbd-is-installapi -emit-tbd -emit-tbd-path %t/flag-provided.tbd+// RUN: %target-swift-frontend -typecheck -o /dev/null %s -tbd-is-installapi -emit-tbd -emit-tbd-path %t/flag-provided.tbd // RUN: %target-swift-frontend -emit-ir -o /dev/null %s -emit-tbd -emit-tbd-path %t/flag-omitted.tbd

I could change the second one, sure. But in -tbd-is-installapi mode, we can’t emit IR anymore.

harlanhaskins

comment created time in an hour

pull request commentapple/swift-corelibs-foundation

[Android][Linux] Fixed hang when using RunLoop.main.run with Date()

@swift-ci please test

Molanda

comment created time in an hour

Pull request review commentapple/foundationdb

Add fdbcli and API command to mark storage servers as permanently failed

 struct RemoveServersSafelyWorkload : TestWorkload {  		state std::vector<ISimulator::ProcessInfo*>	killProcArray; 		state std::vector<AddressExclusion>	toKillArray;-+		state std::vector<AddressExclusion>	toKillMarkFailedArray;+		state AddressExclusion coordExcl;+		// Exclude a coordinator under buggify, but only if fault tolerance is > 0 and kill set is non-empty already+		if (BUGGIFY && toKill.size()) {

If the workload think it is unsafe to kill any servers and we still mark a coordinator as failed without --force , shouldn't the API reject the command because it will fail the safety check?

It is valuable to understand what are the situations we can not kill a coordinator and ensure the safetyCheck can defend against that. This can save a cluster when operators made a mistake in marking coordinators as failed.

fzhjon

comment created time in an hour

startedapple/swift-log

started time in an hour

Pull request review commentapple/swift

[AutoDiff] Add `differentiability_witness_function` instruction.

 bool SILParser::parseSILInstruction(SILBuilder &B) {         InstLoc, extractee, functionOperand);     break;   }+  case SILInstructionKind::DifferentiabilityWitnessFunctionInst: {+    // e.g. differentiability_witness_function+    //      [jvp] [parameters 0 1] [results 0] @foo : $(T) -> T+    DifferentiabilityWitnessFunctionKind witnessKind;+    StringRef witnessKindNames[3] = {"jvp", "vjp", "transpose"};+    SourceLoc lastLoc;+    if (P.parseToken(tok::l_square,+            diag::sil_inst_autodiff_expected_differentiability_witness_kind) ||+        parseSILIdentifierSwitch(witnessKind, witnessKindNames,+            diag::sil_inst_autodiff_expected_differentiability_witness_kind) ||+        P.parseToken(tok::r_square, diag::sil_autodiff_expected_rsquare,+                     "differentiability witness function kind"))+      return true;+    // Parse an index set, prefaced with the given label.+    auto parseIndexSet = [&](StringRef label, SmallVectorImpl<unsigned> &indices,+                             const Diagnostic &parseIndexDiag) -> bool {+      // Parse `[<label> <integer_literal>...]`.+      if (P.parseToken(tok::l_square, diag::sil_autodiff_expected_lsquare,+                       "index list") ||+          P.parseSpecificIdentifier(+              label, diag::sil_autodiff_expected_index_list_label, label))+        return true;+      while (P.Tok.is(tok::integer_literal)) {+        unsigned index;+        if (P.parseUnsignedInteger(index, lastLoc, parseIndexDiag))+          return true;+        indices.push_back(index);+      }+      if (P.parseToken(tok::r_square, diag::sil_autodiff_expected_rsquare,+                       "index list"))+        return true;+      return false;+    };+    // Parse parameter and result indices.+    SmallVector<unsigned, 8> parameterIndices;+    SmallVector<unsigned, 8> resultIndices;+    // Parse parameter and result indices.+    if (parseIndexSet("parameters", parameterIndices,+                      diag::sil_autodiff_expected_parameter_index))+      return true;+    if (parseIndexSet("results", resultIndices,+                      diag::sil_autodiff_expected_result_index))+      return true;+    // Parse a trailing 'where' clause (optional).+    // This represents witness generic signature requirements.+    GenericSignature *witnessGenSig = nullptr;+    SourceLoc whereLoc;+    SmallVector<RequirementRepr, 4> witnessRequirementReprs;+    if (P.Tok.is(tok::l_square) && P.peekToken().is(tok::kw_where)) {+      P.consumeToken(tok::l_square);+      bool firstTypeInComplete;+      P.parseGenericWhereClause(whereLoc, witnessRequirementReprs,+                                firstTypeInComplete,+                                /*AllowLayoutConstraints*/ false);+      if (P.parseToken(tok::r_square, diag::sil_autodiff_expected_rsquare,+                       "where clause"))+        return true;+    }+    // Parse original function name.+    SILFunction *originalFunction;+    if (parseSILFunctionRef(InstLoc, originalFunction))+      return true;+    // Resolve witness requirements.+    if (!witnessRequirementReprs.empty()) {+      // Add original function generic parameters to a local scope so that they+      // can be found during type resolution for `RequirementRepr`s.+      Scope S(&P, ScopeKind::Generics);+      auto *origGenEnv = originalFunction->getGenericEnvironment();+      if (origGenEnv)+        for (auto pd : origGenEnv->getGenericParams())+          P.addToScope(pd->getDecl());+      SmallVector<Requirement, 4> requirements;+      auto *whereClause = TrailingWhereClause::create(+          originalFunction->getModule().getASTContext(), whereLoc,+          witnessRequirementReprs);+      convertRequirements(originalFunction, whereClause->getRequirements(),

Fixed witness generic signature parsing by changing the syntax.

Previously, resolving witness generic requirements based on the original function's type/generic environment was problematic. If the original SILFunction is a bodiless declaration, it will not have a GenericEnvironment, so where clause RequirementRepr resolution will fail. Using the generic environment from the original function type's generic signature (function->getLoweredFunctionType()->getGenericSignature()->getGenericEnvironment()) did not work either.

// Old syntax.
differentiability_witness_function ... [where T: Differentiable] @foo : $(T) -> T

Now, the witness generic signature is parsed separately as a new generic environment with its own generic parameters and requirements:

// New syntax.
differentiability_witness_function ... <T where T: Differentiable> @foo : $(T) -> T

Parsing validation checks that the parsing witness generic signature is compatible with the original function generic signature (e.g. generic parameters are the same), then merges requirements from both to form the final witness generic signature.

dan-zheng

comment created time in an hour

Pull request review commentapple/swift

[TBDGen] Skip non-inlinable function bodies in InstallAPI

  // 1. Emit two TBDs, one with -tbd-is-installapi set, and one without -// RUN: %target-swift-frontend -emit-ir -o /dev/null %s -tbd-is-installapi -emit-tbd -emit-tbd-path %t/flag-provided.tbd+// RUN: %target-swift-frontend -typecheck -o /dev/null %s -tbd-is-installapi -emit-tbd -emit-tbd-path %t/flag-provided.tbd // RUN: %target-swift-frontend -emit-ir -o /dev/null %s -emit-tbd -emit-tbd-path %t/flag-omitted.tbd

Seems weird that these two aren't symmetrical anymore?

harlanhaskins

comment created time in an hour

pull request commentapple/swift

Disable ArrayNew.swift.gyb in armv7k or armv7s

Build failed Swift Test OS X Platform Git Sha - 836b929bd08abd4a040bd6cb4b237bddc10c4817

rintaro

comment created time in an hour

push eventapple/llvm-project-v2-split

Jonas Devlieghere

commit sha ee50d1c3beffa30ba66e7353c80a590ed8ab9e6a

[CMake] Make it possible to set the RPATH in add_lldb_exectable. Make it possible to pass a build and install RPATH to add_lldb_executable instead of having to call lldb_setup_rpaths after the fact. This fixes a real issue where setting an install RPATH with lldb_setup_rpaths would only affect the symroot installation component. Given that lldb_setup_rpaths sets a target property I would expect this to be orthogonal to installation components. Regardless, it makes sense to integrate this functionality in add_lldb_exectable. git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375068 91177308-0d34-0410-b5e6-96231b3b80d8 (cherry picked from commit 0f53352cb22da3e0bd5e4d4272e9849e9883f11e)

view details

swift-ci

commit sha 4e7bd7460062fe9c4fff8209b04afb047f1cf183

Merge pull request #2069 from JDevlieghere/cherrypick/r375068

view details

push time in an hour

push eventapple/llvm-project-v2

Jonas Devlieghere

commit sha 5cda67ea2ce6a77a10965b232c4033cf3f3ced3a

[CMake] Make it possible to set the RPATH in add_lldb_exectable. Make it possible to pass a build and install RPATH to add_lldb_executable instead of having to call lldb_setup_rpaths after the fact. This fixes a real issue where setting an install RPATH with lldb_setup_rpaths would only affect the symroot installation component. Given that lldb_setup_rpaths sets a target property I would expect this to be orthogonal to installation components. Regardless, it makes sense to integrate this functionality in add_lldb_exectable. git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375068 91177308-0d34-0410-b5e6-96231b3b80d8 (cherry picked from commit 0f53352cb22da3e0bd5e4d4272e9849e9883f11e) apple-llvm-split-commit: ee50d1c3beffa30ba66e7353c80a590ed8ab9e6a apple-llvm-split-dir: lldb/

view details

swift-ci

commit sha 37ef0d43d9e2a2e9f664ef92eeed8c4b10c20da7

Merge pull request #2069 from JDevlieghere/cherrypick/r375068 apple-llvm-split-commit: 4e7bd7460062fe9c4fff8209b04afb047f1cf183 apple-llvm-split-dir: lldb/

view details

push time in an hour

push eventapple/llvm-project-v4-split

Jonas Devlieghere

commit sha ee50d1c3beffa30ba66e7353c80a590ed8ab9e6a

[CMake] Make it possible to set the RPATH in add_lldb_exectable. Make it possible to pass a build and install RPATH to add_lldb_executable instead of having to call lldb_setup_rpaths after the fact. This fixes a real issue where setting an install RPATH with lldb_setup_rpaths would only affect the symroot installation component. Given that lldb_setup_rpaths sets a target property I would expect this to be orthogonal to installation components. Regardless, it makes sense to integrate this functionality in add_lldb_exectable. git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375068 91177308-0d34-0410-b5e6-96231b3b80d8 (cherry picked from commit 0f53352cb22da3e0bd5e4d4272e9849e9883f11e)

view details

swift-ci

commit sha 4e7bd7460062fe9c4fff8209b04afb047f1cf183

Merge pull request #2069 from JDevlieghere/cherrypick/r375068

view details

push time in an hour

push eventapple/llvm-project-v4

Jonas Devlieghere

commit sha c937e332f3cef667c2f0a302d789f825bd7bec61

[CMake] Make it possible to set the RPATH in add_lldb_exectable. Make it possible to pass a build and install RPATH to add_lldb_executable instead of having to call lldb_setup_rpaths after the fact. This fixes a real issue where setting an install RPATH with lldb_setup_rpaths would only affect the symroot installation component. Given that lldb_setup_rpaths sets a target property I would expect this to be orthogonal to installation components. Regardless, it makes sense to integrate this functionality in add_lldb_exectable. git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375068 91177308-0d34-0410-b5e6-96231b3b80d8 (cherry picked from commit 0f53352cb22da3e0bd5e4d4272e9849e9883f11e) apple-llvm-split-commit: ee50d1c3beffa30ba66e7353c80a590ed8ab9e6a apple-llvm-split-dir: lldb/

view details

swift-ci

commit sha ef4c28b72dfcab1a5006ee5e326a36565480c480

Merge pull request #2069 from JDevlieghere/cherrypick/r375068 apple-llvm-split-commit: 4e7bd7460062fe9c4fff8209b04afb047f1cf183 apple-llvm-split-dir: lldb/

view details

push time in an hour

Pull request review commentapple/foundationdb

Add fdbcli and API command to mark storage servers as permanently failed

 std::pair<vector<ShardsAffectedByTeamFailure::Team>,vector<ShardsAffectedByTeamF  void ShardsAffectedByTeamFailure::erase(Team team, KeyRange const& range) { 	if(team_shards.erase( std::pair<Team,KeyRange>(team, range) ) > 0) {-		for(auto uid = team.servers.begin(); uid != team.servers.end(); ++uid)-			storageServerShards[*uid]--;+		for (auto uid = team.servers.begin(); uid != team.servers.end(); ++uid) {+			if (storageServerShards[*uid] > 0) {

Can you add a comment here to explain the concern?

fzhjon

comment created time in an hour

Pull request review commentapple/foundationdb

Add fdbcli and API command to mark storage servers as permanently failed

 ACTOR Future<Void> snapCreate(Database cx, Standalone<StringRef> snapCmd, UID sn 		throw; 	} }++ACTOR Future<bool> checkSafeExclusions(Database cx, vector<AddressExclusion> exclusions) {+	TraceEvent("ExclusionSafetyCheckBegin")+	    .detail("NumExclusion", exclusions.size())+	    .detail("Exclusions", describe(exclusions));+	state ExclusionSafetyCheckRequest req(exclusions);+	state bool ddCheck;+	try {+		loop {+			choose {+				when(wait(cx->onMasterProxiesChanged())) {}+				when(ExclusionSafetyCheckReply _ddCheck =+				         wait(loadBalance(cx->getMasterProxies(false), &MasterProxyInterface::exclusionSafetyCheckReq,+				                          req, cx->taskID))) {+					ddCheck = _ddCheck.safe;+					break;+				}+			}+		}+	} catch (Error& e) {+		if (e.code() != error_code_actor_cancelled) {+			TraceEvent("ExclusionSafetyCheckError")+			    .detail("NumExclusion", exclusions.size())+			    .detail("Exclusions", describe(exclusions))+			    .error(e);+		}+		throw;+	}+	TraceEvent("ExclusionSafetyCheckCoordinators");+	state ClientCoordinators coordinatorList(cx->getConnectionFile());+	state vector<Future<Optional<LeaderInfo>>> leaderServers;+	for (int i = 0; i < coordinatorList.clientLeaderServers.size(); i++) {+		leaderServers.push_back(retryBrokenPromise(coordinatorList.clientLeaderServers[i].getLeader,+		                                           GetLeaderRequest(coordinatorList.clusterKey, UID()),+		                                           TaskPriority::CoordinationReply));+	}+	wait(smartQuorum(leaderServers, leaderServers.size() / 2 + 1, 1.5) || delay(2.0));

Can you first resolve other comments in a comment, and try remove this in a new commit? Once you are done, I can run some correctness on both.

fzhjon

comment created time in an hour

Pull request review commentapple/swift

[TBDGen] Mangle symbols before putting them in the TBD

 #include "llvm/ADT/StringRef.h" #include "llvm/ADT/StringSet.h" #include "llvm/IR/Module.h"+#include "llvm/IR/Mangler.h"

Nitpick: the alphabet

harlanhaskins

comment created time in an hour

startedapple/swift

started time in an hour

startedapple/swift-log

started time in an hour

pull request commentapple/swift

Merge branch 'tensorflow' into tensorflow-merge

Oh I thought it's to the tensorflow branch.

asuhan

comment created time in an hour

push eventapple/swift-lldb

Jonas Devlieghere

commit sha ee50d1c3beffa30ba66e7353c80a590ed8ab9e6a

[CMake] Make it possible to set the RPATH in add_lldb_exectable. Make it possible to pass a build and install RPATH to add_lldb_executable instead of having to call lldb_setup_rpaths after the fact. This fixes a real issue where setting an install RPATH with lldb_setup_rpaths would only affect the symroot installation component. Given that lldb_setup_rpaths sets a target property I would expect this to be orthogonal to installation components. Regardless, it makes sense to integrate this functionality in add_lldb_exectable. git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375068 91177308-0d34-0410-b5e6-96231b3b80d8 (cherry picked from commit 0f53352cb22da3e0bd5e4d4272e9849e9883f11e)

view details

swift-ci

commit sha 4e7bd7460062fe9c4fff8209b04afb047f1cf183

Merge pull request #2069 from JDevlieghere/cherrypick/r375068

view details

push time in an hour

PR merged apple/swift-lldb

[CMake] Make it possible to set the RPATH in add_lldb_exectable.

Make it possible to pass a build and install RPATH to add_lldb_executable instead of having to call lldb_setup_rpaths after the fact.

This fixes a real issue where setting an install RPATH with lldb_setup_rpaths would only affect the symroot installation component. Given that lldb_setup_rpaths sets a target property I would expect this to be orthogonal to installation components. Regardless, it makes sense to integrate this functionality in add_lldb_exectable.

git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375068 91177308-0d34-0410-b5e6-96231b3b80d8 (cherry picked from commit 0f53352cb22da3e0bd5e4d4272e9849e9883f11e)

+15 -2

1 comment

1 changed file

JDevlieghere

pr closed time in an hour

pull request commentapple/swift

[reflection] Go back to 0/1 for bitwise_takable instead of false/true

SwiftPM failure? I don't think so.

@swift-ci Please test macOS

jrose-apple

comment created time in an hour

pull request commentapple/swift

[docs] CToSwift: document omit-needless-words

I'm not sure I expect anyone to read this monster document. I'm not sure if it will even come in use. But it does feel like having a prose + pseudocode description of the algorithm is some kind of "understanding redundancy".

@swift-ci Please smoke test

jrose-apple

comment created time in an hour

PR opened apple/swift

Reviewers
[docs] CToSwift: document omit-needless-words

...and update the dirty lie that property names aren't transformed.

The only thing left to do is methods!

+291 -1

0 comment

2 changed files

pr created time in an hour

startedapple/swift

started time in an hour

push eventapple/swift

Victor Guerra

commit sha c76bde94865c608cc21c5b2566f31d14bbc71726

Defines derivatives for remaining tgmath math functions. (#27559) * Inverse trigonometric functions. * Exponents and logarithms. * Hyperbolic functions. * Error functions ( erf, erfc ). * Free generic functions: sqrt, fma. Partially resolves [TF-812](https://bugs.swift.org/browse/TF-812).

view details

Dan Zheng

commit sha 36676ae7181997fcb56114ce00266aebb3c67c84

[AutoDiff] Remap `apply` callee type in derivative context. (#27590) Previously, `LinearMapInfo::addLinearMapToStruct` did not remap `apply` callee type in derivative context. Now, remapping is done. Remapping is significant when the derivative has a more constrained generic signature. Resolves TF-817.

view details

Dan Zheng

commit sha fb6045c22f5c700be32547f26f71d94a6ce86a31

[AutoDiff] Underscore the `Differentiable` protocol. (#27577) Rename the `Differentiable` protocol to `_Differentiable`. This matches upstreaming PR https://github.com/apple/swift/pull/27511, which adds `protocol _Differentiable` in a new `Differentiable.swift` file. `tensorflow` branch defines `typealias Differentiable = _Differentiable` for usability. There should be no user-facing changes, except that some error messages now show '_Differentiable' instead of 'Differentiable'.

view details

Richard Wei

commit sha eeeeee243066e6e19827457d7a075e218a030780

[AutoDiff] Remove differentiation order from AD-related instructions. (#27579) The differentiation order field in `differentiable_function` and `differentiable_function_extract` instructions is unsupported and will not be used by the current design. Quite a lot of dead code exists to try to handle `order`, but it is mostly incomplete and untested. This PR removes the differentiation order from the code base to simplify what we upstream to the 'master' branch. Changes include: * Remove `differentiationOrder` from `DifferentiableFunctionInst` and `DifferentiableFunctionExtractInst`. * Make `DifferentiableFunctionInst::DifferentiableFunctionInst` take an optional pair of JVP and VJP instead of a variable-size array. * Rename "associated functions" to "derivative functions" in `DifferentiableFunctionInst` to align better with [the design](https://forums.swift.org/t/differentiable-programming-mega-proposal/28547). Filed task [TF-882](https://bugs.swift.org/browse/TF-882) to track the renaming of all other occurrences of "associated functions". Resolves [TF-880](https://bugs.swift.org/browse/TF-880).

view details

Gogul Balakrishnan

commit sha 0aee08abee642a0bc51c2091a7a3fa2dec70929a

[AutoDiff] Rename "assocFn" to "derivativeFn" everywhere except Differentiation.cpp. (#27597)

view details

Richard Wei

commit sha 431cc4373423e36dd065345bd67f88a57e6f6c98

[AutoDiff] Rename "associated function" to "derivative function". (#27603) `assocFn` -> `derivativeFn` `AssocFn` -> `DerivativeFn` `assocFunc` -> `derivativeFunc` `AssocFunc` -> `DerivativeFunc` `associatedFunction` -> `derivativeFunction` `AssociatedFunction` -> `DerivativeFunction` `autodiff associated function` -> `derivative function` `autodiff-associated function` -> `derivative function` `AD associated function` -> `derivative function` `associated differentiation function` -> `derivative function` This is a follow-up to #27597. Resolves [TF-882](https://bugs.swift.org/browse/TF-882).

view details

Dan Zheng

commit sha ced939b3008aa17b8787ad3f1d546055bd1128b3

[AutoDiff] Fix subset parameters thunk `partial_apply` substitutions. (#27604) Fix `partial_apply` substitution map for subset parameters linear map thunk. The correct substitution map is computed by `buildThunkType` in the helper `ADContext::getOrCreateSubsetParametersThunkForLinearMap` and is now returned by the helper. Resolves TF-886.

view details

Richard Wei

commit sha 9886b63567a7df169fec9509a8679e920da8dacf

[AutoDiff] Rename 'AutoDiffIndexSubset' to 'IndexSubset'. (#27615) Per @DougGregor's review comment (https://github.com/apple/swift/pull/27555#discussion_r333757205), `AutoDiffIndexSubset` data structure could have a more general name. We rename it to `IndexSubset` and move it to its own file.

view details

Dan Zheng

commit sha 68b96a216a9d0a8382688796338c5a3e89dc5a6e

[AutoDiff] Type-check `@differentiable` attributes during validation. (#27613) Type-check `@differentiable` attributes during `TypeChecker::validateDecl` for all relevant declaration kinds (initializers, subscripts, variables), not just function declarations. Resolves TF-888. TF-789 tracks proper request-based type-checking for `@differentiable` attribute. Exposes TF-892: `ElementaryFunctions` linker error on Linux.

view details

Richard Wei

commit sha 483a37870ee44016089048a5892e2c321aa9505a

[AutoDiff] NFC: Move some 'IndexSubset' method impls to their own file. (#27627) - Move some 'IndexSubset' method implemenetations from AutoDiff.cpp to a new file called IndexSubset.cpp. - Fix some indentation errors caused by the earlier renaming of `AutoDiffIndexSubset`.

view details

Richard Wei

commit sha 7e039731ec64554618f356a00bfd7a753013874e

[AutoDiff] Fix 'autodiff_function_extract' operand ownership kind. (#27638) `ValueOwnershipKindClassifier` and `OperandOwnershipKindClassifier` should have the same classification for `autodiff_function_extract`. `ValueOwnershipKindClassifier`'s classification was fixed by #27199, which gave the correct ownership verification results. Now we fix it in `OperandOwnershipKindClassifier`. `DifferentiableFunctionExtractOriginalExpr`'s SILGen is now corrected to borrowing the argument and emitting a copy for the extracted original function.

view details

Richard Wei

commit sha 65ebee0c7cea75ed757e179b0a9898b0cb4ecde7

[AutoDiff] Fix `@differentiable(linear)` type parsing ambiguity. (#27646) Handle the case where `@differentiable(linear)` is followed by a type alias or another type attribute. ```swift typealias linear = (Float) -> Float let property: @differentiable(linear) linear // okay let property: @differentiable(linear) @convention(c) linear // okay ``` Resolves [TF-576](https://bugs.swift.org/browse/TF-576).

view details

Dan Zheng

commit sha 9c76a294502dc1da556761ffec61f504dafbd675

[AutoDiff] Add `@differentiable` attribute SILGen assertion. (#27650) Previously, `@differentiable` attribute SILGen lowering was skipped if the attribute did not have resolved parameter indices. This led to unexpected and difficult-to-debug bugs like TF-888. Re-add the assertion for robustness. This is possible after https://github.com/apple/swift/pull/27613, which fixed the underlying issue with parameter indices resolution.

view details

Dan Zheng

commit sha a5dc918e702ca3433efe8c362330ea1ca11241a9

[AutoDiff] Add SIL differentiability witnesses. (#27487) SIL differentiability witnesses are a new top-level SIL construct mapping "original" SIL functions to derivative SIL functions. They will replace SIL function `[differentiable]` attributes, additionally enabling cross-module retroactive derivative registration. SIL differentiability witnesses have the following components: - Original `SILFunction`. - Linkage. - Parameter indices (`IndexSubset`). - Result indices (`IndexSubset`). - Derivative generic signature (optional). - JVP `SILFunction` (optional). - VJP `SILFunction` (optional). - "Is serialized?" bit. This patch adds the `SILDifferentiabilityWitness` data structure, along with parsing, printing, verification, and serialization (including lookup by key). The TF-866 master issue tracks follow-up, including SILGen and differentiation transform changes.

view details

Dan Zheng

commit sha 98f35450fafbbb36c6a7c24652f59e7429528125

[NFC] [AutoDiff] Gardening. (#27651)

view details

Richard Wei

commit sha db95e53d2a32214c32c159d9f954eefa57eefb54

[AutoDiff] [ASTGen] Check for 'linear' when generating 'AttributedTypeRepr'. (#27656) ASTGen did not know how to handle `@differentiable(linear)`, which caused parsed `@differentiable(linear)` function types to lose the "linear" bit when they are an argument or a result in a SIL function's type signature. This patch teaches ASTGen to check for `linear` in `TypeAttributes` when generating AST. Resolves [TF-901](https://bugs.swift.org/browse/TF-901).

view details

Richard Wei

commit sha 5e52226a6ab786dd901f690ed91a1b459eccb2b0

[AutoDiff] [Serialization] Fix '@differentiable(linear)' SIL function type serialization. (#27659) Before `DifferentiabilityKind` was introduced, a function was either `@differentiable` or not `@differentiable`. `DifferentiabilityKind` added a third option: `@differentiability(linear)`. SIL function type serialization did not care about the specific `DifferentiabilityKind` but just whether the function is differentiable -- it is now fixed in this patch. Resolves [TF-903](https://bugs.swift.org/browse/TF-903).

view details

Richard Wei

commit sha 4a4fe8308c212d0cea5cf86e4903e5dfcb1704e5

[AutoDiff] [IRGen] Lower `@differentiable(linear)` function types. (#27661) This patch adds support for lowering `@differentiable(linear)` function types to LLVM IR. * Add `SILFunctionType` transpose type calculation utility: `SILFunctionType:: getAutoDiffTransposeFunctionType`. * Refactor `TypeClassifierBase` methods used for computing recursive properties of `@differentiable` function types and eliminated the need for repeatedly checking differentiability kind. * Add `LinearDifferentiableSILFunctionTypeLowering`. Rename the original `DifferentiableSILFunctionTypeLowering` to `NormalDifferentiableSILFunctionTypeLowering` to make clear it's not for linear functions. * Add `TypeConverter::convertLinearDifferentiableFunctionType`. Rename the original `TypeConverter::convertDifferentiableFunctionType` to `TypeConverter::convertNormalDifferentiableFunctionType` to make clear it's not for linear functions. Resolves [TF-902](https://bugs.swift.org/browse/TF-902).

view details

Richard Wei

commit sha 740b63ed2dd3d39c5041e55203949e40058f669f

[AutoDiff] Introduce 'linear_function' and 'linear_function_extract' instructions. (#27637) Introduce `linear_function` and `linear_function_extract` instructions, which are used for creating and destructing `@differentiable(linear)` functions. ### `linear_function` instruction Bundles a function with its transpose function into a `@differentiable(linear)` function. ``` sil-instruction ::= 'linear_function' sil-linear-function-parameter-indices? sil-value ':' sil-type sil-linear-function-transpose-function-clause? sil-linear-function-parameter-indices ::= '[' 'parameters' [0-9]+ (' ' [0-9]+)* ']' sil-linear-transpose-function-clause ::= with_transpose sil-value ':' sil-type ``` ### `linear_function_extract` instruction Extracts the original function or a transpose function from the given ``@differentiable(linear)`` function. It must be provided with an extractee: ``[original]`` or ``[transpose]``. ``` sil-instruction ::= 'linear_function_extract' sil-linear-function-extractee sil-value ':' sil-type sil-linear-function-extractee ::= '[' sil-linear-function-extractee ']' sil-linear-function-extractee-name ::= 'original' | 'transpose' linear_function_extract [original] %0 : $@differentiable(linear) (T) -> T linear_function_extract [transpose] %0 : $@differentiable(linear) (T) -> T ``` Resolves [TF-907](https://bugs.swift.org/browse/TF-907).

view details

Dan Zheng

commit sha bb67311187ff8fe925a349978a60ab4ba0ed0760

[AutoDiff] Diagnose unsupported forward-mode control flow. (#27684) Diagnose unsupported forward-mode control flow instead of crashing.

view details

push time in 2 hours

delete branch apple/swift

delete branch : asuhan/tensorflow-merge

delete time in 2 hours

PR merged apple/swift

Merge branch 'tensorflow' into tensorflow-merge

Bring the new changes in the tensorflow branch.

+5038 -2332

3 comments

135 changed files

asuhan

pr closed time in 2 hours

pull request commentapple/swift

Merge branch 'tensorflow' into tensorflow-merge

I don't believe CI is strictly necessary on merge PRs other than the final apple/tensorflow-merge -> apple/tensorflow PR, so let's perhaps merge this now and create + test the final PR.

asuhan

comment created time in 2 hours

startedapple/darwin-xnu

started time in 2 hours

pull request commentapple/swift

[SIL Opt][OSLogOptimization] Add -enable-ownership-stripping-after-serialization flag

Build failed Swift Test Linux Platform Git Sha - 641cc8b660d623b8c4b3684806102e12c9ead76b

ravikandhadai

comment created time in 2 hours

startedapple/swift-driver

started time in 2 hours

push eventapple/llvm-project-v2

Kousik Kumar

commit sha 962ca076e51c25a7a08f4e0d329c65328a635bdb

Include sanitize blacklist and other extra deps as part of scan-deps output Summary: Clang's -M mode includes these extra dependencies in its output and clang-scan-deps should have equivalent behavior, so adding these extradeps to output just like how its being done for ".d" file generation mode. Reviewers: arphaman, dexonsmith, Bigcheese, jkorous Subscribers: cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D69017 llvm-svn: 375074

view details

push time in 2 hours

push eventapple/llvm-project-v2-split

Kousik Kumar

commit sha 5ab89ea40b058a1e68a38e711db3013a9a057daf

Include sanitize blacklist and other extra deps as part of scan-deps output Summary: Clang's -M mode includes these extra dependencies in its output and clang-scan-deps should have equivalent behavior, so adding these extradeps to output just like how its being done for ".d" file generation mode. Reviewers: arphaman, dexonsmith, Bigcheese, jkorous Subscribers: cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D69017 git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@375074 91177308-0d34-0410-b5e6-96231b3b80d8

view details

push time in 2 hours

push eventapple/llvm-project-v4-split

Kousik Kumar

commit sha 5ab89ea40b058a1e68a38e711db3013a9a057daf

Include sanitize blacklist and other extra deps as part of scan-deps output Summary: Clang's -M mode includes these extra dependencies in its output and clang-scan-deps should have equivalent behavior, so adding these extradeps to output just like how its being done for ".d" file generation mode. Reviewers: arphaman, dexonsmith, Bigcheese, jkorous Subscribers: cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D69017 git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@375074 91177308-0d34-0410-b5e6-96231b3b80d8

view details

push time in 2 hours

push eventapple/llvm-project-v3

Kousik Kumar

commit sha 962ca076e51c25a7a08f4e0d329c65328a635bdb

Include sanitize blacklist and other extra deps as part of scan-deps output Summary: Clang's -M mode includes these extra dependencies in its output and clang-scan-deps should have equivalent behavior, so adding these extradeps to output just like how its being done for ".d" file generation mode. Reviewers: arphaman, dexonsmith, Bigcheese, jkorous Subscribers: cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D69017 llvm-svn: 375074

view details

push time in 2 hours

push eventapple/llvm-project-v4

Kousik Kumar

commit sha 962ca076e51c25a7a08f4e0d329c65328a635bdb

Include sanitize blacklist and other extra deps as part of scan-deps output Summary: Clang's -M mode includes these extra dependencies in its output and clang-scan-deps should have equivalent behavior, so adding these extradeps to output just like how its being done for ".d" file generation mode. Reviewers: arphaman, dexonsmith, Bigcheese, jkorous Subscribers: cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D69017 llvm-svn: 375074

view details

push time in 2 hours

startedapple/swift-driver

started time in 2 hours

startedapple/swift

started time in 2 hours

fork Lynjay/swift

The Swift Programming Language

https://swift.org

fork in 2 hours

pull request commentapple/swift

[TBDGen] Skip non-inlinable function bodies in InstallAPI

@swift-ci please test

harlanhaskins

comment created time in 2 hours

pull request commentapple/swift-syntax

Model SyntaxVisitor as a class

Build failed Swift Test OS X Platform Git Sha - 11a32cf0f7dad164ec0ab2b290a17c724c6ef66a

ahoppen

comment created time in 2 hours

pull request commentapple/swift

[SkipFunctionBodies] Build everything but the stdlib with function body skipping

@swift-ci please test

harlanhaskins

comment created time in 2 hours

pull request commentapple/swift

[SkipFunctionBodies] Build everything but the stdlib with function body skipping

@swift-ci please test debug

harlanhaskins

comment created time in 2 hours

PR opened apple/swift

[SkipFunctionBodies] Build everything but the stdlib with function body skipping

This un-reverts #27588, which was causing failures on debug builds because it was only disabling the NonInlinableFunctionSkippingChecker for the optimized OnoneSupport module, but not the unoptimized one.

+7 -1

0 comment

2 changed files

pr created time in 2 hours

PR opened apple/swift

Reviewers
[TBDGen] Skip non-inlinable function bodies in InstallAPI

For now, until we figure out the right way to present -experimental-skip-non-inlinable-function-bodies, make -tbd-is-installapi imply that, for testing.

+7 -1

0 comment

2 changed files

pr created time in 2 hours

pull request commentapple/swift

[ConstraintSystem] Diagnosing invalid explicit coercion via fix

@xedin Thank you for the review, I'll search for JIRA issues right now :)) They are probably marked as DiagnosticsQol label so I can filter for that?

LucianoPAlmeida

comment created time in 2 hours

pull request commentapple/swift

[TBDGen] Mangle symbols before putting them in the TBD

@swift-ci please smoke test

harlanhaskins

comment created time in 2 hours

Pull request review commentapple/swift

[ConstraintSystem] Diagnosing invalid explicit coercion via fix

 bool GenericArgumentsMismatchFailure::diagnoseAsError() {    Optional<Diag<Type, Type>> diagnostic;   if (path.empty()) {-    assert(isa<AssignExpr>(anchor));-    diagnostic = getDiagnosticFor(CTP_AssignSource);+    if (isa<AssignExpr>(anchor)) {+      diagnostic = getDiagnosticFor(CTP_AssignSource);+    } else if (isa<CoerceExpr>(anchor)) {+      diagnostic = getDiagnosticFor(CTP_CoerceOperand);+    } else {+      assert(false && "unhandled expression type");

Sure, Done :)

LucianoPAlmeida

comment created time in 2 hours

push eventapple/apple-llvm-infrastructure-tools

Duncan P. N. Exon Smith

commit sha dc965943a3d3cd3fa31cf7f26d62c31d883aa942

apple-v5: start wtih apple-v4

view details

Duncan P. N. Exon Smith

commit sha 0e38d9b9384e80fb9e6e0fdcff549e2ebb0d177a

apple-v5: Repeat `dir`s in favour of `repeat`s The `split2mono` tool supports "auto" repeats of `dir`s that are shared between branches. When we can use that it's more precise than a `repeat`, because it will follow `--ancestry-path` instead of relying on `--first-parent` (which could fail to generate a good history). Prefer that where possible. Also fixed two things as drive-bys: - Filled in the start directive for swift/master/start, which had been half typed in before. - Fixed spelling for the clang-tools-extra `dir` on apple/master/root-start.

view details

push time in 2 hours

PR merged apple/apple-llvm-infrastructure-tools

Reviewers
apple-v5: Prepare for another regeneration

We're regenerating (again) after a bugfix in split2mono.

The main change is to repeat dirs in favour of repeats.

The split2mono tool supports "auto" repeats of dirs that are shared between branches. When we can use that it's more precise than a repeat, because it will follow --ancestry-path instead of relying on --first-parent (which could fail to generate a good history). Prefer that where possible.

Also fixed two things as drive-bys:

  • Filled in the start directive for swift/master/start, which had been half typed in before.
  • Fixed spelling for the clang-tools-extra dir on apple/master/root-start.
+338 -0

2 comments

1 changed file

dexonsmith

pr closed time in 2 hours

push eventapple/apple-llvm-infrastructure-tools

Duncan P. N. Exon Smith

commit sha 42316ee780bbb073b5a60d4a0ab40514dd2b07c5

mt-generate: Ensure that `repeat` is a no-op when it should be `mt-translate-ref` is sometimes (but rarely) called when there are new commits on the branch getting repeated, but none of them affect the "repeated" dirs. Ensure that we don't generate merge commits for those, since it makes the history extra confusing (even without this change, the content was correct). As a drive-by, avoid comparing trees twice.

view details

push time in 2 hours

PR merged apple/apple-llvm-infrastructure-tools

Reviewers
mt-generate: Ensure that `repeat` is a no-op when it should be

mt-translate-ref is sometimes (but rarely) called when there are new commits on the branch getting repeated, but none of them affect the "repeated" dirs. Ensure that we don't generate merge commits for those, since it makes the history extra confusing (even without this change, the content was correct).

As a drive-by, avoid comparing trees twice.

+151 -49

5 comments

6 changed files

dexonsmith

pr closed time in 2 hours

PR opened apple/swift

Reviewers
[TBDGen] Mangle symbols before putting them in the TBD

The linker expects to see mangled symbols in the TBD, otherwise it won't be able to link anything. Use LLVM's mangler to mangle them.

Fixes rdar://54055049

+102 -6

0 comment

5 changed files

pr created time in 2 hours

pull request commentapple/apple-llvm-infrastructure-tools

mt-generate: Ensure that `repeat` is a no-op when it should be

@swift-ci please test

dexonsmith

comment created time in 2 hours

pull request commentapple/apple-llvm-infrastructure-tools

mt-generate: Ensure that `repeat` is a no-op when it should be

@swift-ci please test again

dexonsmith

comment created time in 2 hours

pull request commentapple/apple-llvm-infrastructure-tools

mt-generate: Ensure that `repeat` is a no-op when it should be

@swift-ci test again

dexonsmith

comment created time in 2 hours

push eventapple/llvm-project-v2-split

Fangrui Song

commit sha a8e32ad42b065adfb51051e2cd8ef650a5088579

SBCommandReturnObject: change LLDB_RECORD_METHOD(..., FILE *, ...) to use LLDB_RECORD_DUMMY POSIX says FILE is a typedef to a structure containing information about a file. The structure is unspecified, i.e. it may be an incomplete type, as is the case on musl (`struct _IO_FILE` is an implementation detail that is not exposed). `LLDB_RECORD_METHOD(..., (FILE *), ...)` transitively uses sizeof(FILE) and requires the structure to be complete. Change it to LLDB_RECORD_DUMMY to fix the build failure on musl (regression of D57475). Reviewed By: JDevlieghere, labath, lawrence_danna Differential Revision: https://reviews.llvm.org/D68872 git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375072 91177308-0d34-0410-b5e6-96231b3b80d8

view details

Lawrence D'Anna

commit sha f464db9300815dc9263ad74114f91e1197d70336

delete SWIG typemaps for FILE* Summary: The SWIG typemaps for FILE* are no longer used, so this patch deletes them. Reviewers: JDevlieghere, jasonmolenda, labath Reviewed By: labath Subscribers: lldb-commits Tags: #lldb Differential Revision: https://reviews.llvm.org/D68963 git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375073 91177308-0d34-0410-b5e6-96231b3b80d8

view details

swift-ci

commit sha 56eccc89cd2b6c34a945ba41ef7765df7d9ec88d

Merge remote-tracking branch 'llvm/master' into upstream-with-swift

view details

push time in 2 hours

push eventapple/llvm-project-v2

Dmitry Mikulin

commit sha 15984457a673eff3969c0755be543e7d572edd47

Revert Tag CFI-generated data structures with "#pragma clang section" attributes. This reverts r375022 (git commit e2692b3bc0327606748b6d291b9009d2c845ced5) llvm-svn: 375069

view details

Daniel Sanders

commit sha 149a02042531980ffdad0314922bd10544484d70

Fix unused variable in r375066 llvm-svn: 375070

view details

Daniel Sanders

commit sha 6150093e22815d49634beceed0472fddc9ef33fb

Correct placement of #ifndef NDEBUG in r375067 llvm-svn: 375071

view details

Fangrui Song

commit sha 56ee31964f5a57621953eacdc8458bd41dfc4154

SBCommandReturnObject: change LLDB_RECORD_METHOD(..., FILE *, ...) to use LLDB_RECORD_DUMMY POSIX says FILE is a typedef to a structure containing information about a file. The structure is unspecified, i.e. it may be an incomplete type, as is the case on musl (`struct _IO_FILE` is an implementation detail that is not exposed). `LLDB_RECORD_METHOD(..., (FILE *), ...)` transitively uses sizeof(FILE) and requires the structure to be complete. Change it to LLDB_RECORD_DUMMY to fix the build failure on musl (regression of D57475). Reviewed By: JDevlieghere, labath, lawrence_danna Differential Revision: https://reviews.llvm.org/D68872 llvm-svn: 375072

view details

Lawrence D'Anna

commit sha 0f783599a4c645d8ae826f990f7b938fac6e5dae

delete SWIG typemaps for FILE* Summary: The SWIG typemaps for FILE* are no longer used, so this patch deletes them. Reviewers: JDevlieghere, jasonmolenda, labath Reviewed By: labath Subscribers: lldb-commits Tags: #lldb Differential Revision: https://reviews.llvm.org/D68963 llvm-svn: 375073

view details

swift-ci

commit sha e6135e4863b50b988b1cdf8baea9f2c1ff994ef9

Merge remote-tracking branch 'llvm/master' into upstream-with-swift apple-llvm-split-commit: 56eccc89cd2b6c34a945ba41ef7765df7d9ec88d apple-llvm-split-dir: lldb/

view details

push time in 2 hours

push eventapple/llvm-project-v4-split

Fangrui Song

commit sha a8e32ad42b065adfb51051e2cd8ef650a5088579

SBCommandReturnObject: change LLDB_RECORD_METHOD(..., FILE *, ...) to use LLDB_RECORD_DUMMY POSIX says FILE is a typedef to a structure containing information about a file. The structure is unspecified, i.e. it may be an incomplete type, as is the case on musl (`struct _IO_FILE` is an implementation detail that is not exposed). `LLDB_RECORD_METHOD(..., (FILE *), ...)` transitively uses sizeof(FILE) and requires the structure to be complete. Change it to LLDB_RECORD_DUMMY to fix the build failure on musl (regression of D57475). Reviewed By: JDevlieghere, labath, lawrence_danna Differential Revision: https://reviews.llvm.org/D68872 git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375072 91177308-0d34-0410-b5e6-96231b3b80d8

view details

Lawrence D'Anna

commit sha f464db9300815dc9263ad74114f91e1197d70336

delete SWIG typemaps for FILE* Summary: The SWIG typemaps for FILE* are no longer used, so this patch deletes them. Reviewers: JDevlieghere, jasonmolenda, labath Reviewed By: labath Subscribers: lldb-commits Tags: #lldb Differential Revision: https://reviews.llvm.org/D68963 git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375073 91177308-0d34-0410-b5e6-96231b3b80d8

view details

swift-ci

commit sha 56eccc89cd2b6c34a945ba41ef7765df7d9ec88d

Merge remote-tracking branch 'llvm/master' into upstream-with-swift

view details

push time in 2 hours

push eventapple/llvm-project-v4

Dmitry Mikulin

commit sha 15984457a673eff3969c0755be543e7d572edd47

Revert Tag CFI-generated data structures with "#pragma clang section" attributes. This reverts r375022 (git commit e2692b3bc0327606748b6d291b9009d2c845ced5) llvm-svn: 375069

view details

Daniel Sanders

commit sha 149a02042531980ffdad0314922bd10544484d70

Fix unused variable in r375066 llvm-svn: 375070

view details

Daniel Sanders

commit sha 6150093e22815d49634beceed0472fddc9ef33fb

Correct placement of #ifndef NDEBUG in r375067 llvm-svn: 375071

view details

Fangrui Song

commit sha 56ee31964f5a57621953eacdc8458bd41dfc4154

SBCommandReturnObject: change LLDB_RECORD_METHOD(..., FILE *, ...) to use LLDB_RECORD_DUMMY POSIX says FILE is a typedef to a structure containing information about a file. The structure is unspecified, i.e. it may be an incomplete type, as is the case on musl (`struct _IO_FILE` is an implementation detail that is not exposed). `LLDB_RECORD_METHOD(..., (FILE *), ...)` transitively uses sizeof(FILE) and requires the structure to be complete. Change it to LLDB_RECORD_DUMMY to fix the build failure on musl (regression of D57475). Reviewed By: JDevlieghere, labath, lawrence_danna Differential Revision: https://reviews.llvm.org/D68872 llvm-svn: 375072

view details

Lawrence D'Anna

commit sha 0f783599a4c645d8ae826f990f7b938fac6e5dae

delete SWIG typemaps for FILE* Summary: The SWIG typemaps for FILE* are no longer used, so this patch deletes them. Reviewers: JDevlieghere, jasonmolenda, labath Reviewed By: labath Subscribers: lldb-commits Tags: #lldb Differential Revision: https://reviews.llvm.org/D68963 llvm-svn: 375073

view details

swift-ci

commit sha 457502d1d51dbef27573e0744649c83ca58fb2c3

Merge remote-tracking branch 'llvm/master' into upstream-with-swift apple-llvm-split-commit: 56eccc89cd2b6c34a945ba41ef7765df7d9ec88d apple-llvm-split-dir: lldb/

view details

push time in 2 hours

pull request commentapple/swift

[reflection] Go back to 0/1 for bitwise_takable instead of false/true

Build failed Swift Test OS X Platform Git Sha - ac21ee486f98622054cf54296c3e2843c021271a

jrose-apple

comment created time in 2 hours

startedapple/swift

started time in 2 hours

push eventapple/swift-lldb

Fangrui Song

commit sha a8e32ad42b065adfb51051e2cd8ef650a5088579

SBCommandReturnObject: change LLDB_RECORD_METHOD(..., FILE *, ...) to use LLDB_RECORD_DUMMY POSIX says FILE is a typedef to a structure containing information about a file. The structure is unspecified, i.e. it may be an incomplete type, as is the case on musl (`struct _IO_FILE` is an implementation detail that is not exposed). `LLDB_RECORD_METHOD(..., (FILE *), ...)` transitively uses sizeof(FILE) and requires the structure to be complete. Change it to LLDB_RECORD_DUMMY to fix the build failure on musl (regression of D57475). Reviewed By: JDevlieghere, labath, lawrence_danna Differential Revision: https://reviews.llvm.org/D68872 git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375072 91177308-0d34-0410-b5e6-96231b3b80d8

view details

Lawrence D'Anna

commit sha f464db9300815dc9263ad74114f91e1197d70336

delete SWIG typemaps for FILE* Summary: The SWIG typemaps for FILE* are no longer used, so this patch deletes them. Reviewers: JDevlieghere, jasonmolenda, labath Reviewed By: labath Subscribers: lldb-commits Tags: #lldb Differential Revision: https://reviews.llvm.org/D68963 git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375073 91177308-0d34-0410-b5e6-96231b3b80d8

view details

swift-ci

commit sha 56eccc89cd2b6c34a945ba41ef7765df7d9ec88d

Merge remote-tracking branch 'llvm/master' into upstream-with-swift

view details

push time in 2 hours

pull request commentapple/swift

[AutoDiff] Clean up.

Merging to unblock progress.

dan-zheng

comment created time in 2 hours

push eventapple/swift

Dan Zheng

commit sha 92b0f221113ba202f113716b0b245246b8da2cd9

[AutoDiff] Clean up. (#27718) - Store `AutoDiffConfig` in `SILDifferentiabilityWitness` instead of storing the individual components. This makes it cheaper to get an `AutoDiffConfig`. - Unify parsing logic and diagnostics. - Minor style changes.

view details

push time in 2 hours

PR merged apple/swift

[AutoDiff] Clean up. tensorflow
  • Store AutoDiffConfig in SILDifferentiabilityWitness instead of storing the individual components. This makes it cheaper to get an AutoDiffConfig.
  • Unify parsing logic and diagnostics.
  • Minor style changes.
+96 -160

4 comments

8 changed files

dan-zheng

pr closed time in 2 hours

push eventapple/llvm-project-v2

Lawrence D'Anna

commit sha 0f783599a4c645d8ae826f990f7b938fac6e5dae

delete SWIG typemaps for FILE* Summary: The SWIG typemaps for FILE* are no longer used, so this patch deletes them. Reviewers: JDevlieghere, jasonmolenda, labath Reviewed By: labath Subscribers: lldb-commits Tags: #lldb Differential Revision: https://reviews.llvm.org/D68963 llvm-svn: 375073

view details

push time in 2 hours

push eventapple/llvm-project-v2-split

Fangrui Song

commit sha a8e32ad42b065adfb51051e2cd8ef650a5088579

SBCommandReturnObject: change LLDB_RECORD_METHOD(..., FILE *, ...) to use LLDB_RECORD_DUMMY POSIX says FILE is a typedef to a structure containing information about a file. The structure is unspecified, i.e. it may be an incomplete type, as is the case on musl (`struct _IO_FILE` is an implementation detail that is not exposed). `LLDB_RECORD_METHOD(..., (FILE *), ...)` transitively uses sizeof(FILE) and requires the structure to be complete. Change it to LLDB_RECORD_DUMMY to fix the build failure on musl (regression of D57475). Reviewed By: JDevlieghere, labath, lawrence_danna Differential Revision: https://reviews.llvm.org/D68872 git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375072 91177308-0d34-0410-b5e6-96231b3b80d8

view details

Lawrence D'Anna

commit sha f464db9300815dc9263ad74114f91e1197d70336

delete SWIG typemaps for FILE* Summary: The SWIG typemaps for FILE* are no longer used, so this patch deletes them. Reviewers: JDevlieghere, jasonmolenda, labath Reviewed By: labath Subscribers: lldb-commits Tags: #lldb Differential Revision: https://reviews.llvm.org/D68963 git-svn-id: https://llvm.org/svn/llvm-project/lldb/trunk@375073 91177308-0d34-0410-b5e6-96231b3b80d8

view details

push time in 2 hours

more