src/kudu/tablet/concurrent_btree.h (8 lines): - line 347: // TODO: inline slices with more than 8 bytes will store a prefix of the - line 452: // TODO: switch to std::assume_aligned once we reach C++20. - line 1040: // TODO: this call probably won't be necessary in the final implementation - line 1097: // TODO: unit test - line 1392: // TODO: simplified implementation doesn't deal with splitting - line 1395: << "TODO: currently only support splitting nodes with >2 keys"; - line 1397: // TODO: can we share code better between the node types here? - line 1443: // TODO: this could be done more efficiently since we know that src/kudu/cfile/rle_block.h (6 lines): - line 61: // TODO (perf) : doing this one bit a time is probably - line 182: // TODO : do this a word/byte at a time as opposed bit at a time - line 222: // TODO : consider if this can also be used for BOOL with only minor - line 305: // TODO : as with the matching BlockBuilder above (see comments for - line 372: // TODO (perf): make this faster by moving forward a 'run at a time' - line 374: // TODO (perf): investigate placing pointers somewhere in either the java/kudu-client/src/main/java/org/apache/kudu/client/Bytes.java (5 lines): - line 603: // TODO: Support larger/smaller than 16 bytes (int128) - line 629: // TODO: Support larger/smaller than 16 bytes (int128) - line 649: // TODO: Support larger/smaller than 16 bytes (int128) - line 836: // TODO: use n.unscaledValue().intValueExact() when we drop Java7 support. - line 844: // TODO: use n.unscaledValue().intValueExact() when we drop Java7 support. src/kudu/master/catalog_manager.cc (5 lines): - line 918: // TODO (KUDU-591): This is a bit of a hack, as right now - line 1412: // TODO: Abdicate on timeout instead of crashing. - line 1415: << " election. TODO: Abdicate instead."; - line 4736: // TODO: if there is no replica available, should we still keep the task running? - line 4738: // TODO: don't pick replica we can't lookup??? src/kudu/client/batcher.cc (5 lines): - line 159: // TODO: not implemented yet - line 169: // TODO: not implemented yet - line 533: // TODO: This error handling block should really be rewritten to handle - line 919: // TODO Keep a replica picker per tablet and share it across writes - line 939: // TODO: there is a potential race here -- if the Batcher gets destructed while src/kudu/common/row.h (4 lines): - line 69: // TODO: potential optimization here: if the new value is smaller than - line 84: memcpy(dst->mutable_ptr(), src.ptr(), src.size()); // TODO: inline? - line 283: // TODO: doesn't look like this uses the is_identity performance - line 297: // TODO: Discourage the use of this. At the moment is only in RowChangeList::Project src/kudu/common/partial_row.cc (4 lines): - line 135: // TODO: at some point we could allow type coercion here. - line 506: // TODO We can probably remove this when we move to c++11 and can use "extern template" - line 841: // TODO: at some point we could allow type coercion here. - line 866: // TODO: allow specifying a prefix of the key, and automatically src/kudu/consensus/consensus_queue.cc (4 lines): - line 380: // TODO: we should probably refactor the ResponseFromPeer function - line 413: // TODO: it would be a cleaner design to explicitly set the first index in the - line 1292: // calculation. See the TODO in AdvanceQueueWatermark() for more details. - line 1573: // TODO: we should probably use an atomic here. We'll just annotate src/kudu/tserver/tablet_service.cc (4 lines): - line 2599: // TODO: add test case for this invalid request - line 2669: // TODO: remove this once all clients have moved to ColumnPredicatePB and - line 2987: // TODO: This validation essentially prohibits scans with READ_AT_SNAPSHOT - line 3163: // TODO: should check if RPC got cancelled, once we implement RPC cancellation. src/kudu/consensus/consensus.proto (4 lines): - line 189: // TODO: At the time of writing, this field is never set in the response. - line 190: // TODO: Propagate signed timestamps. See KUDU-611. - line 325: // TODO: Migrate ConsensusService to the AppStatusPB RPC style and merge these errors. - line 547: // TODO: Consider renaming these to copy_source_*. java/kudu-client/src/main/java/org/apache/kudu/client/Operation.java (3 lines): - line 349: // TODO: would be more efficient to use a buffer which "chains" smaller allocations - line 363: // TODO: we could avoid a copy here by using an implementation that allows - line 369: // TODO: same as above, we could avoid a copy here by using an implementation that allows src/kudu/cfile/encoding-test.cc (3 lines): - line 893: // TODO: Uncomment when adding 128 bit support to RLE (KUDU-2284) - line 906: // TODO: Uncomment when adding 128 bit support to RLE (KUDU-2284) - line 919: // TODO: Uncomment when adding 128 bit support to RLE (KUDU-2284) src/kudu/tablet/tablet_mm_ops.cc (3 lines): - line 183: // TODO: we should acquire the rowset compaction locks here. Otherwise, until - line 269: // TODO: See CompactRowSetsOp::Prepare(). - line 357: // TODO: See CompactRowSetsOp::Prepare(). src/kudu/scripts/backup-perf.py (3 lines): - line 108: # TODO: Change this to use the Kudu python API. - line 145: # TODO: This may no longer be needed if and when we integrate - line 169: # TODO: Non-string columns are assumed to be 8 bytes. src/kudu/gutil/thread_annotations.h (3 lines): - line 135: #define NO_THREAD_SAFETY_ANALYSIS_FIXME \ - line 154: // TS_FIXME is used to mark lock expressions that are not valid C++ syntax. - line 225: #define NO_THREAD_SAFETY_ANALYSIS_FIXME \ src/kudu/cfile/cfile_reader.h (3 lines): - line 162: // delta reader code. TODO: think about reorganizing this: - line 242: // TODO: do we ever want to be able to seek to the end of the file? - line 338: // TODO: do we ever want to be able to seek to the end of the file? src/kudu/tablet/memrowset.cc (3 lines): - line 201: // TODO: for now, the key ends up stored doubly -- - line 500: // TODO: add dcheck that dst->schema() matches our schema - line 501: // also above TODO applies to a lot of other CopyNextRows cases src/kudu/util/rle-encoding.h (3 lines): - line 61: // TODO: think about how to use this for strings. The bit packing isn't quite the same. - line 147: // TODO: consider adding a min_repeated_run_length so the caller can control - line 203: // TODO: generalize this to other bit widths src/kudu/cfile/block_encodings.h (3 lines): - line 116: // TODO: Since we know the actual file position, maybe we - line 180: // TODO: change this to a Remaining() call? - line 191: // TODO: get rid of this from the block decoder, and put it in a generic src/kudu/benchmarks/tpch/line_item_tsv_importer.h (3 lines): - line 90: // TODO: extra copy here, since we don't have a way to parse StringPiece - line 103: // TODO: extra copy here, since we don't have a way to parse StringPiece - line 116: // TODO: extra copy here, since we don't have a way to parse StringPiece src/kudu/consensus/raft_consensus.cc (3 lines): - line 1283: // TODO: this isn't just an optimization! if we comment this out, we get - line 1318: // TODO move this to raft_consensus-state or whatever we transform that into. - line 1465: // TODO - These failure scenarios need to be exercised in an unit java/kudu-client/src/main/java/org/apache/kudu/client/AsyncKuduScanner.java (3 lines): - line 464: * TODO: Remove the need for this by handling server side. - line 852: return client.closeScanner(this).addCallback(closedCallback()); // TODO errBack ? - line 1342: // TODO: Find a clean way to plumb in reuseRowResult. java/kudu-client/src/main/java/org/apache/kudu/client/AsyncKuduClient.java (3 lines): - line 1545: // TODO maybe we can retry it? - line 2257: // TODO we don't always need to sleep, maybe another replica can serve this RPC. - line 2379: // TODO: If the TS advertises multiple host/ports, pick the right one src/kudu/util/inline_slice.h (2 lines): - line 42: // TODO: we could store a prefix of the indirect data in this unused space - line 118: // TODO: if already indirect and the current storage has enough space, just reuse that. java/kudu-backup/src/main/scala/org/apache/kudu/backup/Options.scala (2 lines): - line 54: // TODO (KUDU-1260): Add a test and enable by default? - line 94: // TODO (KUDU-2677): Document the limitations based on cluster configuration. src/kudu/master/master.proto (2 lines): - line 364: // TODO: perhaps add some very basic metrics reporting here, like - line 441: // TODO: new clients should prefer the 'Interned' type below. src/kudu/scripts/dump_breakpad_symbols.py (2 lines): - line 99: TODO: Lookup the binary in the system path. Not urgent, since the user can specify the - line 106: # TODO: Use dump_syms_mac if on macOS. java/kudu-client/src/main/java/org/apache/kudu/client/Negotiator.java (2 lines): - line 363: // TODO: check that the message type matches the expected one in all - line 1035: * TODO (ghenke): Remove the need for this reflection. src/kudu/server/server_base.cc (2 lines): - line 798: // TODO: maybe actually bump a sequence number on local disk instead of - line 1159: // TODO: Use PB container format? java/kudu-backup/src/main/scala/org/apache/kudu/backup/KuduRestore.scala (2 lines): - line 107: // TODO (KUDU-2788): Store the full metadata to compare/validate for each applied partial. - line 226: // TODO (KUDU-2832): If the job fails to restore a table it may still create the table, which src/kudu/client/meta_cache.cc (2 lines): - line 168: // TODO: if the TS advertises multiple host/ports, pick the right one - line 632: // TODO if we add new Pick* methods the method to (re-)call needs to be passed as src/kudu/common/wire_protocol.cc (2 lines): - line 494: // TODO: add test case for this invalid request - line 1043: // TODO: Using LLVM to build a specialized CopyColumn on the fly should have src/kudu/codegen/code_generator.cc (2 lines): - line 173: // TODO: this may fail if there are multiple 'ret' instructions in one - line 206: // TODO would be nice to just initialize the TargetMachine here, just once, src/kudu/common/common.proto (2 lines): - line 105: // TODO: Differentiate between the schema attributes - line 268: // TODO implement actually signing the propagated timestamp. src/kudu/client/client.cc (2 lines): - line 1459: // TODO: implement me - right now be conservative. - line 2177: // TODO: do some double-buffering here -- when we return this batch src/kudu/tablet/compaction.cc (2 lines): - line 1719: // TODO: on this pass, we don't actually need the row data, just the - line 1847: // TODO: there should be a more elegant way of preventing metadata flush at this point src/kudu/codegen/code_cache.h (2 lines): - line 49: // TODO: currently CodeCache is implemented using the Cache in - line 56: // TODO: a potential improvment would be for the cache to monitor its memory src/kudu/tablet/delta_tracker.cc (2 lines): - line 686: // TODO perf: is this really needed? Will check - line 698: // TODO: we need to somehow ensure this doesn't fail - we have to somehow coordinate src/kudu/gutil/linux_syscall_support.h (2 lines): - line 3966: * TODO: Test whether ARM needs a restorer - line 4481: // TODO: define this in an arch-independant way instead of inlining the clone src/kudu/cfile/plain_bitmap_block.h (2 lines): - line 60: // TODO (perf) : doing this one bit a time is probably - line 183: // TODO : do this a word/byte at a time as opposed bit at a time src/kudu/client/predicate-test.cc (2 lines): - line 261: // TODO: uncomment after fixing KUDU-1386 - line 288: // TODO: uncomment after fixing KUDU-1386 src/kudu/gutil/int128.h (2 lines): - line 31: // TODO: division, etc. - line 88: // TODO: perhaps it would be nice to have int128, a signed 128-bit type? src/kudu/tserver/tserver_admin.proto (2 lines): - line 138: // TODO: Replace with the table descriptor - line 151: // TODO KUDU-611 propagate timestamps with server signature. python/kudu/client.pyx (2 lines): - line 962: # TODO: is this cheaply knowable? - line 1957: # TODO: boundscheck src/kudu/rpc/serialization.cc (2 lines): - line 183: buf[kHeaderPosServiceClass] = 0; // TODO: implement - line 184: buf[kHeaderPosAuthProto] = 0; // TODO: implement src/kudu/util/compression/compression_codec.cc (2 lines): - line 200: * TODO: use a instance-local Arena and pass alloc/free into zlib - line 222: // TODO: use z_stream src/kudu/util/thread.cc (2 lines): - line 215: // TODO: Add start-time. - line 216: // TODO: Track fragment ID. src/kudu/util/hdr_histogram.cc (2 lines): - line 231: // portion, which gives us a direct index into the sub-bucket. TODO: Right?? - line 546: // TODO: Can this expression be simplified? src/kudu/consensus/log-test.cc (2 lines): - line 973: // TODO: should consider adding batching here of multiple replicates - line 1140: // TODO: We don't currently do bookkeeping to ensure that we check if the src/kudu/tablet/tablet-pushdown-test.cc (2 lines): - line 215: // TODO: support non-key predicate pushdown on columns which aren't - line 219: // TODO: collect IO statistics per column, verify that most of the string blocks src/kudu/tserver/tserver.proto (2 lines): - line 160: // TODO crypto sign this and propagate the signature along with - line 194: // TODO KUDU-611 propagate timestamps with server signature. src/kudu/util/debug/trace_event_impl.cc (2 lines): - line 400: // TODO: it would be even faster to avoid the memory barrier here entirely, - line 704: // FIXME: base/json/json_writer.cc is using the same code, src/kudu/tablet/delta_compaction.h (2 lines): - line 59: // TODO: is base_schema supposed to be the same as base_data->schema()? how about - line 111: // TODO: doc me src/kudu/util/env_posix.cc (2 lines): - line 1093: // TODO (perf) investigate zeroing a pre-allocated allocated area in - line 1762: // TODO: lint: Consider using readdir_r(...) instead of readdir(...) for improved thread safety. java/kudu-backup-common/src/main/scala/org/apache/kudu/backup/TableMetadata.scala (2 lines): - line 355: .toString // TODO: Explicitly control print format - line 375: case Type.DECIMAL => new BigDecimal(value) // TODO: Explicitly pass scale build-support/build_source_release.py (1 line): - line 147: # TODO: Cache and call the jar from the maven repo? src/kudu/rpc/retriable_rpc.h (1 line): - line 152: // TODO Remove this and pass the used replica around. For now we need to keep this as java/kudu-client/src/main/java/org/apache/kudu/util/SecurityUtil.java (1 line): - line 89: // TODO: should we offer some kind of "renewal thread" or src/kudu/rpc/rpc.h (1 line): - line 86: // TODO Currently this only picks the leader, though it wouldn't be unfeasible to have this src/kudu/hms/hive_metastore.thrift (1 line): - line 780: // TODO: we might want to add "bestEffort" flag; where a subset can fail src/kudu/tablet/diskrowset.cc (1 line): - line 216: // TODO: performance might be better if we actually batch this - src/kudu/tools/tool_action_common.cc (1 line): - line 398: // TODO (KUDU-515): Handle the case when a tablet's schema changes src/kudu/client/value.cc (1 line): - line 223: // TODO: Coerce when possible src/kudu/util/metrics.h (1 line): - line 199: // TODO: Output to HTML. src/kudu/common/row_operations-test.cc (1 line): - line 472: // TODO: evaluate whether this should be true - how "dumb" should clients be? src/kudu/cfile/type_encodings.cc (1 line): - line 215: // TODO: Add 128 bit support to RLE src/kudu/consensus/log_cache.cc (1 line): - line 185: // TODO: we should also try to evict from other tablets - probably better to src/kudu/tablet/lock_manager.cc (1 line): - line 578: // TODO: would be nice to hook in some histogram metric about lock acquisition java/kudu-spark-tools/src/main/scala/org/apache/kudu/spark/tools/DistributedDataGenerator.scala (1 line): - line 96: // TODO: Use context.writeRows while still tracking inserts/collisions. src/kudu/rpc/outbound_call.cc (1 line): - line 331: // TODO: here we're deserializing the call response within the reactor thread, src/kudu/util/jwt-util.cc (1 line): - line 635: // TODO support CurlAuthType by calling kudu::EasyCurl::set_auth(). java/kudu-spark/src/main/scala/org/apache/kudu/spark/kudu/KuduReadOptions.scala (1 line): - line 62: val defaultBatchSize: Int = 1024 * 1024 * 20 // TODO: Understand/doc this setting? java/kudu-client/src/main/java/org/apache/kudu/client/KuduTable.java (1 line): - line 354: // TODO: This could be moved into the RangeSchemaPB returned from server src/kudu/util/os-util.h (1 line): - line 35: // TODO: Get stats for parent process. java/kudu-backup-tools/src/main/scala/org/apache/kudu/backup/KuduBackupCleaner.scala (1 line): - line 60: // TODO: This is temporary. It will change when pattern support is added. src/kudu/tablet/all_types-scan-correctness-test.cc (1 line): - line 738: // TODO: Uncomment when adding 128 bit support to RLE (KUDU-2284) java/kudu-client/src/main/java/org/apache/kudu/client/KeyEncoder.java (1 line): - line 172: // TODO find a way to not have to read byte-by-byte that doesn't require extra copies. This is src/kudu/client/scan_token-internal.cc (1 line): - line 203: // TODO: for statements like "create table as select *", auto_incrementing_id has to be src/kudu/util/pb_util.cc (1 line): - line 764: // TODO: Flush just the dirty bytes. python/kudu/schema.pyx (1 line): - line 817: # TODO: std::map is slightly verbose and inefficient here (O(lg n) python/kudu/libkudu_client.pxd (1 line): - line 738: # TODO: Will need to decide on a strategy for exposing the session's src/kudu/tablet/diskrowset-test.cc (1 line): - line 105: // TODO: add test which calls CopyNextRows on an iterator with no more src/kudu/codegen/module_builder.cc (1 line): - line 176: // TODO: consider parsing this module once instead of on each invocation. src/kudu/codegen/code_generator.h (1 line): - line 73: // TODO static ObjectCache shared b/w engines src/kudu/util/atomic-test.cc (1 line): - line 33: // TODO Add some multi-threaded tests; currently AtomicInt is just a src/kudu/tablet/delta_compaction.cc (1 line): - line 81: // TODO: can you major-delta-compact a new column after an alter table in order src/kudu/client/schema.cc (1 line): - line 670: // TODO: eventually allow primary keys which aren't the first column java/kudu-client/src/main/java/org/apache/kudu/client/AsyncKuduSession.java (1 line): - line 390: // TODO: this should be something different than NotFound so that build-support/parse_test_failure.py (1 line): - line 170: # TODO: there is actually some info on the previous line but there src/kudu/client/scanner-internal.h (1 line): - line 274: // TODO: This and the overall scan retry logic duplicates much of RpcRetrier. src/kudu/rpc/result_tracker.h (1 line): - line 381: // TODO consider a per-ClientState lock if we find this too coarse grained. src/kudu/client/scan_batch.cc (1 line): - line 265: // TODO We can probably remove this when we move to c++11 and can use "extern template" java/buildSrc/src/main/groovy/org/apache/kudu/gradle/DistTestTask.java (1 line): - line 87: // TODO: this requires a glob like **/*Foo* instead of just *Foo* src/kudu/codegen/module_builder.h (1 line): - line 78: // TODO: with multiple *.ll files, each file should be loaded on demand src/kudu/tablet/ops/write_op.cc (1 line): - line 297: // FIXME: Since this is called as a void in a thread-pool callback, src/kudu/gutil/sysinfo.cc (1 line): - line 416: cpuinfo_cycles_per_second = EstimateCyclesPerSecond(500); // TODO <500? java/kudu-client/src/main/java/org/apache/kudu/client/KuduRpc.java (1 line): - line 238: * TODO make this cover most if not all RPCs (right now only scans and writes use this). src/kudu/rpc/rpc_introspection.proto (1 line): - line 64: // TODO: swap out for separate fields src/kudu/server/tracing_path_handlers.cc (1 line): - line 252: // occurred. TODO: change the JS to bubble up the actual error message src/kudu/util/group_varint-inl.h (1 line): - line 155: // TODO: remove this and just automatically pick the right implementation at runtime. src/kudu/common/rowblock.h (1 line): - line 306: // NOTE: TODO: We don't have any separate class for ConstRowBlock and ConstColumnBlock src/kudu/util/crc.cc (1 line): - line 34: // TODO: Is initial = 0 and roll window = 4 appropriate for all cases? src/kudu/common/key_encoder.h (1 line): - line 179: // TODO: optimize for the case where the input slice has '\0' src/kudu/client/scanner-internal.cc (1 line): - line 601: // TODO: should ensure that sleep_ms does not pass the provided deadline. src/kudu/util/subprocess.cc (1 line): - line 395: // TODO: prctl(PR_SET_PDEATHSIG) is Linux-specific, look into portable ways src/kudu/tablet/mt-diskrowset-test.cc (1 line): - line 115: // TODO: test that updates were successful -- collect the updated src/kudu/common/schema.h (1 line): - line 940: // ...but with a different type, (TODO: try with an adaptor) src/kudu/util/thread.h (1 line): - line 52: // TODO: would be nice to offer a way to use ptrace() or signals to src/kudu/tablet/mt-tablet-test.cc (1 line): - line 144: // TODO: add a test where some of the inserts actually conflict java/kudu-client/src/main/java/org/apache/kudu/util/AsyncUtil.java (1 line): - line 26: * TODO (KUDU-602): Some of these methods could eventually be contributed back to async or to a src/kudu/benchmarks/tpch/tpch1.cc (1 line): - line 235: // TODO those two are missing at the moment, might want to change Result src/kudu/util/status.h (1 line): - line 496: // TODO: Move error codes into an error_code.proto or something similar. src/kudu/master/sys_catalog.cc (1 line): - line 259: // TODO: In this case we probably should execute the migration step. src/kudu/rpc/rpc-test.cc (1 line): - line 1302: // TODO: Implement an incoming queue latency test. src/kudu/tablet/compaction.h (1 line): - line 109: // TODO: can we make the above less messy? src/kudu/util/monotime-test.cc (1 line): - line 142: // TODO: Reliably test MonoDelta::FromSeconds() considering floating-point rounding errors src/kudu/master/catalog_manager.h (1 line): - line 1504: // (TODO: this stuff should be deferred and done in the background thread) src/kudu/util/bloom_filter.h (1 line): - line 48: // TODO: an SSE CRC32 hash is probably ~20% faster. Come back to this java/config/spotbugs/excludeFilter.xml (1 line): - line 131: src/kudu/cfile/binary_prefix_block.cc (1 line): - line 311: // TODO: Seek calls should return a Status src/kudu/client/client-internal.cc (1 line): - line 302: // TODO: write a proper async version of this for async client. src/kudu/common/rowid.h (1 line): - line 34: // TODO: Currently we only support up to 4B rows per RowSet - some work src/kudu/rpc/rpc_header.proto (1 line): - line 357: // TODO: Make code required? src/kudu/util/bit-util.h (1 line): - line 27: // TODO: is this in boost or something else like that? java/kudu-spark/src/main/scala/org/apache/kudu/spark/kudu/KuduContext.scala (1 line): - line 485: // TODO: At some point we may want to support more or less tasks while still partitioning. src/kudu/tablet/deltafile.cc (1 line): - line 402: // TODO: this does an allocation - can we stack-allocate the bitmap java/kudu-client/src/main/java/org/apache/kudu/client/ConnectToCluster.java (1 line): - line 103: // TODO: Handle the situation when multiple in-flight RPCs all want to query the masters, src/kudu/tablet/memrowset.h (1 line): - line 222: // (TODO: Different schema are not yet supported) src/kudu/rpc/service_if.cc (1 line): - line 43: // TODO remove this once we have fully cluster-tested this. src/kudu/tablet/tablet_random_access-test.cc (1 line): - line 107: // TODO: should add a version of this test which also tries invalid operations src/kudu/cfile/binary_dict_block.cc (1 line): - line 265: // TODO: implement CopyNextAndEval for more blocks. Eg. other blocks can src/kudu/consensus/consensus_peers.cc (1 line): - line 628: // TODO generalize this exponential backoff algorithm, as we do the src/kudu/common/row_changelist.h (1 line): - line 63: // If type == kReinsert, then a "tuple-format" row follows. TODO: this will src/kudu/tablet/rowset.cc (1 line): - line 232: << "TODO: should make sure this is 64-bit safe - probably not right now" java/kudu-jepsen/src/main/clojure/jepsen/kudu.clj (1 line): - line 56: ;; TODO collect table data for debugging src/kudu/util/kernel_stack_watchdog.h (1 line): - line 251: // TODO: technically this barrier is stronger than we need: we are the only writer src/kudu/client/batcher.h (1 line): - line 84: // may time out before even sending an op). TODO: implement that src/kudu/util/env.cc (1 line): - line 53: // TODO: move these utils into env_util src/kudu/common/column_predicate.h (1 line): - line 391: // TODO: this could be improved with a histogram of expected values. src/kudu/util/random.h (1 line): - line 50: // FIXME: This currently only generates 31 bits of randomness. src/kudu/tablet/delta_iterator_merger.cc (1 line): - line 98: // TODO: do we need to do some kind of sorting here to deal with out-of-order src/kudu/tablet/rowset_tree.h (1 line): - line 117: // TODO map to usage statistics as well. See KUDU-??? src/kudu/master/placement_policy.cc (1 line): - line 57: // TODO (oclarms): get the number of times this tablet server has recently been www/metrics-epoch.js (1 line): - line 23: // TODO allow setting this from the UI, e.g., add a pick-and-choose java/kudu-client/src/main/java/org/apache/kudu/client/PartialRow.java (1 line): - line 795: // TODO: use Utf8.isWellFormed from Guava 16 to verify that. src/kudu/consensus/log.cc (1 line): - line 664: // TODO (perf) zero the new segments -- this could result in src/kudu/tablet/ops/op_driver.cc (1 line): - line 430: // TODO: it's a bit silly that we have three copies of the opid: src/kudu/fs/fs.proto (1 line): - line 78: // TODO: add a "node type" (TS/Master?) src/kudu/common/wire_protocol.h (1 line): - line 199: // TODO: would be nice to just return a vector, but src/kudu/tablet/ops/alter_schema_op.cc (1 line): - line 179: // to release this lock as a workaround for KUDU-915. See the TODO in src/kudu/tserver/tablet_replica_lookup.h (1 line): - line 46: // TODO: Rename this interface. src/kudu/util/knapsack_solver.h (1 line): - line 142: std::vector item_taken_; // TODO: record difference vectors? src/kudu/master/placement_policy.h (1 line): - line 154: // TODO (aserbin): add the reference to the document once it's in the repo. src/kudu/tserver/tserver_service.proto (1 line): - line 53: // TODO: Consider refactoring this as a scan that runs a checksum aggregation java/kudu-spark-tools/src/main/scala/org/apache/kudu/spark/tools/ImportExportFiles.scala (1 line): - line 33: @InterfaceStability.Unstable //TODO: Unstable due to KUDU-2454 src/kudu/cfile/cfile_reader.cc (1 line): - line 1145: // TODO: we could add a mark/reset like interface in BlockDecoder interface java/kudu-client/src/main/java/org/apache/kudu/client/RpcProxy.java (1 line): - line 409: // TODO: This is a crutch until we either don't have to retry RPCs going to the java/kudu-client/src/main/java/org/apache/kudu/util/BloomFilter.java (1 line): - line 53: * // TODO: implement the interface for serializing and sending src/kudu/fs/fs_manager-test.cc (1 line): - line 234: // future after the TODO "The new tenant should have its own dd manager instead of sharing" in src/kudu/common/iterator.h (1 line): - line 50: // TODO: passing NULL is just convenience for unit tests, etc. src/kudu/cfile/cfile.proto (1 line): - line 64: // TODO: name all the PBs with *PB convention src/kudu/tablet/metadata.proto (1 line): - line 146: // TODO: maybe this should be TableSchemaPB? Need to actually put those attributes