Summary: 111 instances, 108 unique

Text	Count
buf[kHeaderPosServiceClass] = 0; // TODO: implement	1
// TODO: we should also try to evict from other tablets - probably better to	1
# TODO: Use dump_syms_mac if on macOS.	1
// TODO - explain this more	1
// TODO: is this in boost or something else like that?	1
// TODO: this is pretty lock-heavy. Would be good to sub out something	1
// FIXME: base/json/json_writer.cc is using the same code,	1
// TODO: perhaps it would be nice to have int128, a signed 128-bit type?	1
// TODO: At the time of writing, this field is never set in the response.	1
// TODO Remove this and pass the used replica around. For now we need to keep this as	1
// TODO: disable detector after a few attempts	1
// TODO: add a "node type" (TS/Master?)	1
// TODO (perf) zero the new segments -- this could result in	1
std::vector<bool> item_taken_; // TODO: record difference vectors?	1
// TODO: technically this barrier is stronger than we need: we are the only writer	1
// TODO - Verify if first check is redundant	1
// TODO Add some multi-threaded tests; currently AtomicInt is just a	1
// TODO: increment a metric here on allocation failure.	1
// TODO (perf) investigate zeroing a pre-allocated allocated area in	1
// TODO: Output to HTML.	1
// TODO: Multi hop proxy support needs better failure and health checks	1
// TS_FIXME is used to mark lock expressions that are not valid C++ syntax.	1
// TODO: use z_stream	1
// TODO: it would seem that it would be good to loop around and see if	1
// TODO: an SSE CRC32 hash is probably ~20% faster. Come back to this	1
// TODO: Differentiate between the schema attributes	1
// TODO: upgrade to protobuf 2.5.x and attach service comments	1
// TODO: use the actual RPC name instead of the request type name.	1
// TODO: if already indirect and the current storage has enough space, just reuse that.	1
// TODO move this to raft_consensus-state or whatever we transform that into.	1
// TODO: include port numbers, etc.	2
// TODO: The method used to check for the proxy peer's health only works on	1
// TODO: Track fragment ID.	1
TODO: Lookup the binary in the system path. Not urgent, since the user can specify the	1
// TODO (vinay): Refactor this to not have to copy pointers again into	1
// TODO: Reliably test MonoDelta::FromSeconds() considering floating-point rounding errors	1
// TODO: Migrate ConsensusService to the AppStatusPB RPC style and merge these errors.	1
// TODO: Implement an incoming queue latency test.	1
// calculation. See the TODO in AdvanceQueueWatermark() for more details.	1
// TODO: We don't currently do bookkeeping to ensure that we check if the	1
# TODO: there is actually some info on the previous line but there	1
// TODO: remove this and just automatically pick the right implementation at runtime.	1
// TODO: generalize this to other bit widths	1
// TODO: we should probably use an atomic here. We'll just annotate	1
#define NO_THREAD_SAFETY_ANALYSIS_FIXME \	2
// TODO: should consider adding batching here of multiple replicates	1
// TODO: here we're deserializing the call response within the reactor thread,	1
// TODO generalize this exponential backoff algorithm, as we do the	1
// TODO We currently optimize for freeing log retention but we could consider having some sort of	1
// TODO - iRitwik ( please refine these mechanisms to a standard way of	1
// TODO: Use a proper thread pool implementation.	1
// TODO: consider pushing this onto another thread since it may be slow.	1
// TODO: add test case for this invalid request	1
// TODO: Is initial = 0 and roll window = 4 appropriate for all cases?	1
cpuinfo_cycles_per_second = EstimateCyclesPerSecond(500); // TODO <500?	1
// TODO implement actually signing the propagated timestamp.	1
# TODO: Cache and call the jar from the maven repo?	1
// TODO: validate additional header flags:	1
// TODO: consider adding a min_repeated_run_length so the caller can control	1
// TODO: Get stats for parent process.	1
// TODO consider a per-ClientState lock if we find this too coarse grained.	1
// TODO: it would be a cleaner design to explicitly set the first index in the	1
// TODO: division, etc.	1
// TODO: provide a way to override designated per-region proxy peer	1
// TODO: move these utils into env_util	1
buf[kHeaderPosAuthProto] = 0; // TODO: implement	1
// TODO: Make code required?	1
// FIXME: This currently only generates 31 bits of randomness.	1
// TODO: this isn't just an optimization! if we comment this out, we get	1
// TODO: needs perf testing and maybe implement streaming (un)compression	1
// TODO (anirbanr-fb) - This needs to be evaluated from the standpoint	1
// TODO: Can this expression be simplified?	1
// TODO - These failure scenarios need to be exercised in an unit	1
// TODO: shutdown? probably, since any future stuff on this socket will be	1
// TODO - for now, its not clear which one is more safe.	1
// TODO: Test coverage needs to be improved a lot.	1
// TODO Currently this only picks the leader, though it wouldn't be unfeasible to have this	1
// TODO: lint: Consider using readdir_r(...) instead of readdir(...) for improved thread safety.	1
// TODO -	1
// TODO: Consider renaming these to copy_source_*.	1
// TODO: unordered_map is incompatible with pthread_t, but would be more	1
// TODO: Propagate signed timestamps. See KUDU-611.	1
// TODO: Explain percentile_ticks_per_half_distance.	1
// TODO: we should probably refactor the ResponseFromPeer function	1
// TODO allow setting this from the UI, e.g., add a pick-and-choose	1
// TODO: would be nice to just return a vector<ConstContiguousRow>, but	1
// portion, which gives us a direct index into the sub-bucket. TODO: Right??	1
// TODO: maybe actually bump a sequence number on local disk instead of	1
// TODO: Use PB container format?	1
// TODO: it would be even faster to avoid the memory barrier here entirely,	1
// TODO: Refactor this code. Ideally all fields in 'request_' related to	1
// TODO: swap out for separate fields	1
// TODO remove this once we have fully cluster-tested this.	1
// TODO: Move error codes into an error_code.proto or something similar.	1
// TODO: Add start-time.	1
* TODO: Test whether ARM needs a restorer	1
// TODO: would be better to cancel the transfer while it is still on the queue if we	1
// TODO: would be nice to offer a way to use ptrace() or signals to	1
// TODO what to do about leap seconds? see KUDU-146	1
// TODO: implement	2
// TODO Consider merging this with ScanRpcStatus.	1
//     TODO: we could store a prefix of the indirect data in this unused space	1
// TODO: Using LLVM to build a specialized CopyColumn on the fly should have	1
// TODO: Needs perf testing and maybe add support for streaming compression	1
// TODO: think about how to use this for strings.  The bit packing isn't quite the same.	1
* TODO: use a instance-local Arena and pass alloc/free into zlib	1
// TODO: prctl(PR_SET_PDEATHSIG) is Linux-specific, look into portable ways	1
// TODO: Flush just the dirty bytes.	1