* Use Ubuntu 18.04 and LLVM 9 on Travis.
* Fix bash conditionals: [ a == b ] should be [ a = b ].

PiperOrigin-RevId: 271898719

PiperOrigin-RevId: 268265314

PiperOrigin-RevId: 266001777

A recent change (4cb80b7ddc) to DBImpl::DeleteObsoleteFiles
unlocked DBImpl::mutex_ while deleting files to allow for
greater concurrency. This change improves on the prior in
a few areas:

1. The table is evicted from the table cache before unlocking
   the mutex. This should only improve performance.
2. This implementation is slightly simpler, but at the cost of
   a bit more memory usage.
3. A comment adding more detail as to why the mutex is being
   unlocked and why it is safe to do so.

PiperOrigin-RevId: 253111645

The documentation (README.md and index.md) referred to the
public headers using an incorrect path - fixing.

PiperOrigin-RevId: 252922925

This change was submitted in https://github.com/google/leveldb/pull/575
by @prajwalchalla.

This fixes issue #523.

PiperOrigin-RevId: 252912613

Fixes #695.

PiperOrigin-RevId: 252895299

PiperOrigin-RevId: 252842234

PiperOrigin-RevId: 250702492

PiperOrigin-RevId: 250383036

PiperOrigin-RevId: 250309603

PiperOrigin-RevId: 249899128

Fixes issue #684.

PiperOrigin-RevId: 249531001

mutex_ was already acquired before accessing DBImpl::versions_ in all
but one place: DBImpl::GetApproximateSizes. This change requires mutex_
to be held before accessing versions_.

PiperOrigin-RevId: 248390814

Converted `for (;<condition>;)` to `while (<condition>)`.

PiperOrigin-RevId: 247950510

PiperOrigin-RevId: 247491163

PiperOrigin-RevId: 247483339

PiperOrigin-RevId: 247424040

Switching from snprintf to std::ostringstream eliminates
cast warning for (unsigned long long).

PiperOrigin-RevId: 247326681

PiperOrigin-RevId: 247261470

The previous attempt of having EnvPosix use O_CLOEXEC (close-on-exec()) when opening file descriptors added tests that relied on procfs, which is Linux-specific. These tests failed on macOS. Unfortunately, the test failures were not caught due to a (since fixed) error in our Travis CI configuration.

This CL re-structures the tests to only rely on POSIX features. Since there is no POSIX-compliant way to get a file name/path out of a file descriptor, this CL breaks up the O_CLOEXEC test into multiple tests, where each Env method that creates an FD gets its own test. This is intended to make it easier to find and fix errors in Env implementations.

This CL also fixes the implementation of NewLogger() to use O_CLOEXEC on macOS. The current implementation passes "we" to fopen(), but the macOS standard C library does not implement the "e" flag yet.

PiperOrigin-RevId: 247088953

Two minor corrections to correct the 900f7d37eb commit
to conform to the Google C++ style guide.

PiperOrigin-RevId: 246907647

PiperOrigin-RevId: 246903086

1) Inline EncodeFixed{32,64}(). They emit single machine instructions on 64-bit processors.
2) Remove size narrowing compiler warnings from DecodeFixed{32,64}().
3) Add comments explaining the current state of optimizations in compilers we care about.
4) Change C-style includes, like <stdint.h>, to C++ style, like <cstdint>.
5) memcpy -> std::memcpy.

The optimization comments are based on https://godbolt.org/z/RdIqS1. The missed optimization opportunities in clang have been reported as https://bugs.llvm.org/show_bug.cgi?id=41761

The change does not have significant impact on benchmarks. Results below.

LevelDB:    version 1.22
Date:       Mon May  6 10:42:18 2019
CPU:        72 * Intel(R) Xeon(R) Gold 6154 CPU @ 3.00GHz
CPUCache:   25344 KB
Keys:       16 bytes each
Values:     100 bytes each (50 bytes after compression)
Entries:    1000000
RawSize:    110.6 MB (estimated)
FileSize:   62.9 MB (estimated)

With change
------------------------------------------------
fillseq      :       2.327 micros/op;   47.5 MB/s
fillsync     :    4185.526 micros/op;    0.0 MB/s (1000 ops)
fillrandom   :       3.662 micros/op;   30.2 MB/s
overwrite    :       4.261 micros/op;   26.0 MB/s
readrandom   :       4.239 micros/op; (1000000 of 1000000 found)
readrandom   :       3.649 micros/op; (1000000 of 1000000 found)
readseq      :       0.174 micros/op;  636.7 MB/s
readreverse  :       0.271 micros/op;  408.7 MB/s
compact      :  570495.000 micros/op;
readrandom   :       2.735 micros/op; (1000000 of 1000000 found)
readseq      :       0.118 micros/op;  937.3 MB/s
readreverse  :       0.190 micros/op;  583.7 MB/s
fill100K     :     860.164 micros/op;  110.9 MB/s (1000 ops)
crc32c       :       1.131 micros/op; 3455.2 MB/s (4K per op)
snappycomp   :       3.034 micros/op; 1287.5 MB/s (output: 55.1%)
snappyuncomp :       0.544 micros/op; 7176.0 MB/s

Baseline
------------------------------------------------
fillseq      :       2.365 micros/op;   46.8 MB/s
fillsync     :    4240.165 micros/op;    0.0 MB/s (1000 ops)
fillrandom   :       3.244 micros/op;   34.1 MB/s
overwrite    :       4.153 micros/op;   26.6 MB/s
readrandom   :       4.698 micros/op; (1000000 of 1000000 found)
readrandom   :       4.065 micros/op; (1000000 of 1000000 found)
readseq      :       0.192 micros/op;  576.3 MB/s
readreverse  :       0.286 micros/op;  386.7 MB/s
compact      :  635979.000 micros/op;
readrandom   :       3.264 micros/op; (1000000 of 1000000 found)
readseq      :       0.169 micros/op;  652.8 MB/s
readreverse  :       0.213 micros/op;  519.5 MB/s
fill100K     :    1055.367 micros/op;   90.4 MB/s (1000 ops)
crc32c       :       1.353 micros/op; 2887.3 MB/s (4K per op)
snappycomp   :       3.036 micros/op; 1286.7 MB/s (output: 55.1%)
snappyuncomp :       0.540 micros/op; 7238.6 MB/s
PiperOrigin-RevId: 246856811

Avoids a use before initialization error. This fixes issue #676.

PiperOrigin-RevId: 246855204

PiperOrigin-RevId: 246848402

Allow the use c.h on projects with -Wstrict-prototypes
Modify CMakelist to include -Wstrict-prototypes

Currently, the benchmark used to assess leveldb changes lives in db/. The codebase also contains two benchmarks against other database engines in doc/bench/. Moving all the benchmarks in one place opens up the way for extracting common code.

PiperOrigin-RevId: 246737541

PiperOrigin-RevId: 246680419

1) Convert iterator-based for loops to C++11 foreach loops.
2) Convert "void operator=" to "T& operator=".
3) Switch from copy operators from private to public deleted.
4) Switch from empty ctors / dtors to "= default" where appropriate.

PiperOrigin-RevId: 246679195

PiperOrigin-RevId: 246668103

PiperOrigin-RevId: 246591372

PiperOrigin-RevId: 246558281

1. Correct the class/struct declaration order to be IAW
   the Google C++ style guide[1].
2. For non-copyable classes, switched from non-implemented
   private methods to explicitly deleted[2] methods.
3. Minor const and member initialization fixes.

[1] https://google.github.io/styleguide/cppguide.html#Declaration_Order
[2] http://eel.is/c++draft/dcl.fct.def.delete

PiperOrigin-RevId: 246521844

Easier to read on sites supporting Markdown (i.e. GitHub).

PiperOrigin-RevId: 246385089

Use clang-format to correct formatting to be in agreement with the [Google C++ Style Guide](https://google.github.io/styleguide/cppguide.html). Doing this simplifies the process of accepting changes. Also fixed a few warnings flagged by clang-tidy.

PiperOrigin-RevId: 246350737

The Travis configuration:
1) Installs recent versions of clang and GCC.
2) Sets up the environment so that CMake picks up the installed
   compilers. Previously, the pre-installed clang compiler was used
   instead.
3) Requests a modern macOS image that has all the headers needed by GCC.

The CL also removes now-unnecessary old workarounds from the
Travis configuration.

PiperOrigin-RevId: 245831188

When iter_ is pointing to current key, we can safely move to the next
key to avoid checking current key, which is of course not necessary.

Benchmark shows that 'readseq' has about 8% performance improvement.

Without patch:

>./db_bench --benchmarks=readseq --num=$((4<<20)) --db=/tmp/db --use_existing_db=1
LevelDB:    version 1.21
Date:       Thu Apr 25 09:37:21 2019
CPU:        32 * Intel(R) Xeon(R) CPU E5-2620 v4 @ 2.10GHz
CPUCache:   20480 KB
Keys:       16 bytes each
Values:     100 bytes each (50 bytes after compression)
Entries:    4194304
RawSize:    464.0 MB (estimated)
FileSize:   264.0 MB (estimated)
------------------------------------------------
readseq      :       0.196 micros/op;  565.7 MB/s

With patch:

>./db_bench --benchmarks=readseq --num=$((4<<20)) --db=/tmp/db --use_existing_db=1
LevelDB:    version 1.21
Date:       Thu Apr 25 09:38:20 2019
CPU:        32 * Intel(R) Xeon(R) CPU E5-2620 v4 @ 2.10GHz
CPUCache:   20480 KB
Keys:       16 bytes each
Values:     100 bytes each (50 bytes after compression)
Entries:    4194304
RawSize:    464.0 MB (estimated)
FileSize:   264.0 MB (estimated)
------------------------------------------------
readseq      :       0.181 micros/op;  612.3 MB/s

Signed-off-by: Kyle Zhang <kyle@smartx.com>

Windows linking visibility in shared libraries requires that inner classes are
specifically exported as visible, even if the containing class is exported.

PiperOrigin-RevId: 244886019

Fix a possible array bounds offset issue flagged in
issue #668. Not the source of any known bug, but will
silence any static analyzers.

PiperOrigin-RevId: 243697659

This makes the syntax highlighting a little nicer on GitHub.

PiperOrigin-RevId: 243426806