Summary: 287 instances, 246 unique

Text	Count
// TODO: rate limit reading headers from blob files.	1
// TODO: consider rename to GetSliceHash32	1
# TODO re-enable pipelined write. Not well tested atm	1
&& defined(__GNUC__) /* TODO: IBM XL */	1
// TODO: support WBWI::SingleDelete() with timestamp.	1
// TODO:(gzh) maybe use direct reads/writes here if possible	1
// TODO: maybe handle the tracing status?	1
// TODO: consider rename to Hash32	1
# TODO: enable this once we figure out how to adjust kill odds for WAL-	1
// TODO: introduce perf counters for block cache insertions	1
// FIXME ^^^: there should be no reason for Get() to depend on current	1
// TODO: Do some refactoring and use only one recovery_error_	1
// TODO: rate limit reading footers from blob files.	2
// TODO: Pick files with max_timestamp > trim_ts by each file's timestamp meta	1
// TODO akanksha: Update the condition when asynchronous prefetching is	1
// TODO: Should this error be ignored?	1
// TODO: refactor this to have a better signature, consolidate	1
// TODO (sagar0): Modify AlignedBuffer.Append to allow doing a memmove	1
// TODO remember the iterator is invalidated because of prefix	1
// TODO: This does not follow symbolic links at this point	1
// TODO: What do we do if this returns an error?	2
// FIXME ^^^: there should be no reason for MultiGet() to depend on current	1
// TODO (yanqin) maybe support max_open_files != -1 by creating hard links	1
// TODO (yanqin): make the following check optional, especially when data	1
// TODO: rate limit old blob DB file reads.	1
// the additional set. (TODO: how this can mitigate scalability and	1
// TODO: think about interaction with Merge. If a user key cannot	1
// TODO: maybe check the return value of Close.	1
// TODO: pre-create kTsMax.	1
// TODO: need to unset flush reason?	1
// TODO: Open file or create hard link to prevent the file from being	1
// TODO: Tune the buffer size.	1
# TODO: Make this work with find_package and/or get rid of it	1
// TODO: This is very similar to FindNextUserEntry() and MergeValuesNewToOld().	1
// TODO: explore combining in a struct	1
// TODO: consider changing to Slice	1
prefix_size_ = 4;  // TODO: support different prefix_size	1
// TODO find a better way to pass compaction_options_fifo.	1
// TODO: Refactor code so that BlockType can determine both the C++ type	1
// TODO: a better and complete implementation is needed to ensure strict	1
// TODO: Cleanup io_status in BuildTable and table builders	1
// TODO: introduce dedicated tickers/statistics/counters	3
// TODO cover transaction DB is not covered in this fault test too.	1
// TODO (PR7798).  We should only add the file to the FileManager if it	1
// TODO: John	1
// TODO (yanqin) try SetSnapshotOnNextOperation(). We currently need to take	1
// TODO: refactor Interleaved*Query so that queries can be "prepared" by	1
// TODO: Add verify_table()	1
// TODO: Presently there is no way to differentiate between error/corruption	1
// TODO: Hook the "name" up to the actual Name() of the MergeOperators?	1
// TODO: distinguish between MANIFEST write and CURRENT renaming	4
TODO:	1
// TODO: Implement the trimming in flush code path.	1
// TODO: support WBWI::Put() with timestamp.	1
# TODO: there is such a thing as transactions with WAL disabled. We should	1
// TODO: re-use one top-level index iterator	1
// TODO: handle timestamp corruption like in general iterator semantics	1
// TODO: support timestamp-based conflict checking.	1
// TODO: support larger timestamp sizes	1
//**TODO: What should we do if we failed to	1
// TODO akanksha: Merge this function with TryReadFromCache once async	1
/* TODO: update */	1
// The following is grossly complicated. TODO: clean it up	1
// handled. TODO: allow Bloom checks where max_covering_tombstone_seq==0	1
// TODO: Should handle this error?	1
// TODO: add support for decoding blob indexes in ldb as well	1
// TODO: Break up into multiple records to reduce memory usage on recovery?	1
s.PermitUncheckedError();  //**TODO: What to do on error?	2
// thread_local is part of C++11 and later (TODO: clean up this define)	2
// of a single linear system. (TODO: implement)	1
.PermitUncheckedError();  //**TODO: What do to on error?	1
// TODO: this is a temporarily solution as it is safe but not optimal for	1
// TODO should have error handling though not much we can do...	1
// TODO: we have to open default CF, because of an implementation limitation,	1
// TODO: introduce aggregate (not per-level) block cache miss count	1
// TODO: log error	3
// TODO: Perform trimming before inserting into memtable during recovery.	1
// TODO (haoyu): We only support Get for now. We need to extend the tracing	1
// TODO read record's till the first no corrupt entry?	1
// TODO: Once GetLiveFilesMetaData supports blob files, update the logic	1
// TODO the while loop inherits from two-level-iterator. We don't know	2
// TODO: will be used with InterleavedSolutionStorage?	1
// TODO: Avoid the snapshot stripe map lookup in CompactionRangeDelAggregator	1
// TODO: We can optimize this if we steal 3 bits.  1 bit: this node is	1
// TODO: maybe handle the tracing status?	5
//**TODO: If/When the DBOptions has a registry in it, the ConfigOptions	1
// TODO:  a better way to set or clean the retryable IO error which	1
// TODO: Ideally we want to verify the hash entry	1
// TODO: give log file and sst file different options (log	1
s.PermitUncheckedError();  //**TODO: What to do on error?	1
// TODO: break this include loop	1
//**TODO: Make the simulate fs something that can be loaded	1
// (TODO: APIs to help choose parameters) One option for fallback in	1
// TODO: update per-level perfcontext user_key_return_count for kMerge	1
// TODO: log error	1
// TODO: Should we check for errors here?	1
// TODO (yanqin) support snapshot.	2
// TODO (yanqin) investigate whether we should abort ingestion or	1
s.PermitUncheckedError();  //**TODO: What to do on error?	1
// TODO akanksha: Add perf_times etc.	1
// TODO: consider memory usage of the FilterBitsReader	1
// TODO: Fix those issues so that the Status	1
// TODO (PR7798).  We should only add the file to the FileManager if it	1
// TODO: consider moving ReadOptions from ArenaWrappedDBIter to DBIter to	1
// XXX/FIXME: This is just basic, naive handling of range tombstones,	1
// TODO (yanqin).	1
// TODO (yanqin) find a suitable status code.	1
// TODO: Should handle status here?	1
// TODO: effectively use the existing checksum of the data being writing to	1
// TODO (yanqin) support separating primary index and secondary index in	1
// TODO fix return value/type	1
// TODO (yanqin)	4
//  (TODO: consider using class toku::txnid_set. The reason for using STL	1
// TODO (yanqin) add stats for other cases?	1
// TODO maybe cache the computation result	1
// TODO: we will support WAL tailing soon.	1
# TODO: enable write-prepared	1
// FIXME: code duplication with GetFastLocalBloomBuilderWithContext	1
// TODO: Support counter batch update for partitioned index and	2
// TODO: Should we return an error if we cannot delete the directory?	1
// TODO: Check for unlock error	1
// FIXME: is changed prefix_extractor handled anywhere for hash index?	1
// block before setting up cache keys. TODO: consider setting up a bootstrap	1
// TODO (yanqin): parallelize jobs with threads.	1
// TODO: Should we check for an error here?	1
// TODO: we need to check the cache dump format version and RocksDB version	1
// TODO: min_log_number_to_keep_2pc check needed?	1
// TODO: examine the behavior for corrupted key	2
// TODO: verify checksum	1
// TODO (yanqin) investigate whether we should sync the closed logs for	1
// TODO: introduce perf counter for compression dictionary hit count	1
// TODO: Should the insert error be ignored?	1
// TODO (AP) support indirect buffers, though probably via a less efficient code path	1
// TODO krad: Evaluate if we need to move to a more strict mode where we	1
// TODO: what are the X-functions? Xcalloc, Xrealloc?	1
// TODO: This particular case seems confusing and unnecessary to pin the	2
// TODO akankshamahajan: Update Poll API to take into account min_completions	1
// TODO: Check for error here?	1
// TODO: update options_file_number_ needed?	1
// TODO: Should we check for errors here?	1
// TODO: pre-create kTsMax.	1
// TODO: sub_compact.io_status is not checked like status. Not sure if thats	1
// TODO: this use of operator bool on `tracer_` can avoid unnecessary lock	2
// TODO: maybe handle the tracing status?	1
// TODO: this use of operator bool on `tracer_` can avoid unnecessary lock	1
// TODO: optimize this performance	1
// TODO: rate limit file reads for checksum calculation during file ingestion.	1
// TODO: What if this fails?	1
// TODO: stopwatch DB_GET needed?, perf timer needed?	1
// TODO: The following is duplicated with Cleanup().	1
// TODO: kManifestWriteNoWAL and kFlushNoWAL are misleading. Refactor is	1
// TODO (AP) add keysArray[i].arrayOffset() if the buffer is indirect	1
// TODO: consider rename to SliceHasher32	1
// TODO: make sure the buff is large enough	1
// FIXME: is changed prefix_extractor handled anywhere for hash index?	1
// Some legacy testing stuff  TODO: carefully clean up obsolete parts	1
// TODO (yanqin) maybe account for file metadata bytes for exact accuracy?	1
// TODO: need to be improved since it sort of defeats the purpose of the rate	2
// TODO: change builder to take the option struct	1
// TODO remove this restriction	1
// TODO: Should handle status here?	1
// TODO: new_checksums: to update files to latest file checksum algorithm	1
// TODO: remove this line when options are used in the loader	1
// TODO: this use of operator bool on `tracer_` can avoid unnecessary lock	1
// TODO: rate limit plain table reads.	1
// TODO: temperature, file_checksum, file_checksum_func_name	1
// TODO: update per-level perfcontext user_key_return_count for kMerge	1
// TODO: some things moved toku_instrumentation.h, not necessarily the best	1
// TODO: iterating over all column families under db mutex.	1
// TODO: rate limit file reads for checksum calculation during file	1
// TODO: Write buffer size passed in should be max of all CF's instead	1
// TODO: What to do if we cannot delete the directory?	1
// TODO (yanqin) with a probability, we can use either forward or backward	1
// TODO future: checksum_func for populating checksums	1
// TODO akanksha:: Dedup below code by calling	1
// TODO akanksha: Update TEST_SYNC_POINT after new tests are added.	1
s.PermitUncheckedError();  // TODO: Check the status	1
// TODO (AP) because in that case we have to pass the array directly,	1
// TODO (yanqin) maybe handle the case in which column_families have	1
// TODO: implement batched interface to plain table bloom	1
// TODO offset passed in is not accurate for parallel compression case	2
// TODO: We should introduce a way to explicitly disable verification	1
// NOTE/TODO: We hope to revise this requirement in the future.	1
// TODO: consider optimizations such as	1
// TODO: the name is currently not stored persistently and thus	1
// TODO: verify/validate	1
//   // TODO: configuring Homogeneous Ribbon for arbitrarily large filters	1
// TODO: Check for errors from OnAddFile?	1
// TODO support multi paths?	1
// TODO Also check the IO status when create the Iterator.	1
// TODO: This may incorrectly select small readahead in case partitioned	1
// TODO: consider using expected_values_dir instead, but this is more	1
// TODO: support timestamp	2
// TODO: more details on trade-offs and practical issues.	1
// TODO: Should we check for an error here?	1
// TODO (yanqin) parallelize if key space is large	2
// TODO: implement batched interface to full block reader	1
// TODO what if one drops a column family while transaction(s) still have	1
s.PermitUncheckedError();  // TODO: What should we do with this error?	1
//**TODO: Should this be error be returned or swallowed?	1
options.sample_for_compression),  // TODO: is 0 fine here?	1
// TODO: source location info might have to be pulled up one caller	1
// TODO: introduce perf counters for misses per block type	1
// TODO: maybe handle the tracing status?	1
// TODO: support timestamp	8
// FIXME: should be a parameter for reading table properties to use persistent	1
// TODO: Re-style this comment to be like the first one	1
// TODO: Don't ignore errors from allocate	1
// TODO: rate limit reads of whole cuckoo tables.	1
// TODO: range unlock does nothing...	1
// TODO (yanqin) allow user to configure probability of each operation.	1
// TODO: easier config for bloom (maybe based on avg key/value size)	1
// TODO: introduce dedicated perf counter for range tombstones	1
s.PermitUncheckedError();  // TODO: What should we do on error?	3
// TODO this experimental option isn't made configurable	1
// TODO: add perf counter for compression dictionary read time	1
// TODO: consider fixed-column specializations with stack-allocated state	1
escalation). (TODO: it is not clear why these operations are tracked with	1
// TODO: rate limit reading headers from blob files.	1
MULTIPLIER);  // TODO: add large key support	1
// TODO: kManifestWriteNoWAL and kFlushNoWAL are misleading. Refactor	1
// TODO with incremental compaction is supported, we might want to	1
// TODO: simplify using GetRefedColumnFamilySet?	1
# TODO: May need to adjust random odds if kill_random_test	1
// TODO: rate limit `BlobLogSequentialReader` reads (it appears unused?)	1
// TODO: move it to different files, as it's testing an internal API	2
// TODO: This mutex should be removed later, to improve performance when	4
// TODO: Add better error handling.	1
// TODO: rate limit reading footers from blob files.	1
// TODO: Need to pass appropriate deadline to TryReadFromCache(). Right now,	1
// TODO: consider rename to LegacyBloomHash32	1
// TODO (yanqin) enable compaction filter	1
// TODO: much more to come	1
// FIXME: can be inconsistent with DisableFileDeletions in cases like	1
// TODO: Check for an error here	1
// TODO: Set status for individual keys appropriately	1
// TODO: consider not counting these as Bloom hits to more closely	1
// TODO: Implement ReadOnly MultiGet?	1
// TODO: a bug here. This function actually does not necessarily	1
// TODO: In the future, BackgroundErrorReason will only be used to indicate	1
// TODO: rate limit footer reads.	1
// TODO (AP) support indirect buffers	1
// TODO: support WBWI::Delete() with timestamp.	1
# TODO detect a hanging condition. The job might run too long as RocksDB	1
// TODO: process future tags such as checksum.	1
* FIXME: Clang's output is still _much_ faster -- On an AMD Ryzen 3600,	1