[
https://issues.apache.org/jira/browse/ARROW-1473?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16191720#comment-16191720
]
ASF GitHub Bot commented on ARROW-1473:
---------------------------------------
GitHub user siddharthteotia opened a pull request:
https://github.com/apache/arrow/pull/1163
ARROW-1473: [JAVA] Initial prototype code hierarchy for review
cc @jacques-n , @BryanCutler , @icexelloss , @elahrvivaz
1. **BaseFixedWidthVector abstract class** - can be extended with a
template to generate all the fixed width vector types.
2. **FixedValueVectorsPrototype.java** - template similar to what we have
currently in FixedValueVectors.java . However, all the common functionality has
been moved from the template to BaseFixedWidthVector. So the template pretty
much contains the custom mutator and accessor methods which are extremely
complex as there is no partial approach to refactor them.
This is why I was suggesting that we should try to get rid of template for
fixed value vectors.
Another approach could be to use the template for simple types like INT,
FLOAT4, FLOAT8 since the "if" conditions in template code are probably not very
complicated for these types. But for types like TIMESTAMP and others, there are
giant "if" blocks.
3. **PrototypeIntVectorNonCodegen** - Non code generated IntVector class
that implements IntVector specific functionality by extending
BaseFixedWidthVector. The get() functions in the Accessor bypass ArrowBuf and
directly work with the memory address to fish out the value.
4. **BaseVariableWidthVector abstract class** - looking at the code of
VarCharVector and VarBinaryVector, it seems like the code is 95% same with only
1 or 2 accessor functions being different. So we can implement all the
functionality (including mutator and accessor as well) in the base class and
then have small non code generated subclasses that just have the specific
functionality for VarChar and VarBinary. We will no longer need
VariableLengthVectors.java template.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/siddharthteotia/arrow ARROW-1473
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/arrow/pull/1163.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #1163
----
commit b62f99a603a4822756c610e52e4643980d4bebf1
Author: Phillip Cloud <[email protected]>
Date: 2017-09-28T00:58:08Z
ARROW-1607: [C++] Implement DictionaryBuilder for Decimals
Author: Phillip Cloud <[email protected]>
Closes #1128 from cpcloud/ARROW-1607 and squashes the following commits:
3137566 [Phillip Cloud] ARROW-1607: [C++] Implement DictionaryBuilder for
Decimals
commit 78181d8ecfce3e658e143210adfc658d6259dec0
Author: Philipp Moritz <[email protected]>
Date: 2017-09-28T13:46:02Z
ARROW-1609: [Plasma] Xcode 9 compilation workaround
see https://github.com/apache/arrow/pull/1139
Author: Philipp Moritz <[email protected]>
Closes #1144 from pcmoritz/plasma-xcode-9-workaround and squashes the
following commits:
7642499 [Philipp Moritz] fix on other platforms
c5cdddd [Philipp Moritz] xcode 9 compilation workaround
commit bcb29d081f26852c2d7b8329f542b1e81bfc9681
Author: Uwe L. Korn <[email protected]>
Date: 2017-09-28T13:47:11Z
ARROW-1620: Python: Download Boost in manylinux1 build from bintray
Author: Uwe L. Korn <[email protected]>
Closes #1141 from xhochy/ARROW-1620 and squashes the following commits:
30da182 [Uwe L. Korn] ARROW-1620: Python: Download Boost in manylinux1
build from bintray
commit 7323677665c5b71203a1e526c0a19eb5520442d0
Author: siddharth <[email protected]>
Date: 2017-09-29T13:58:09Z
ARROW-1618: [JAVA] Reduce Heap Usage (Phase 1)
cc @jacques-n , @icexelloss , @BryanCutler
This is initial small phase of our attempt to reduce heap usage per vector.
As part of investigation, we realized its better to address few things as
part of subtasks for ARROW-1463. Accordingly, I need to update the requirements
document w.r.t heap usage for ARROW-1471.
This patch gets rid of Release Listener object in Allocation Manager as all
the logic is implemented as part of AllocationManager itself.
https://docs.google.com/document/d/1MU-ah_bBHIxXNrd7SkwewGCOOexkXJ7cgKaCis5f-PI/edit
Author: siddharth <[email protected]>
Closes #1142 from siddharthteotia/ARROW-1618 and squashes the following
commits:
77151a27 [siddharth] ARROW-1618: Reduce Heap Usage (Phase 1)
commit f8cf91d25eed5aebebb4aaec61fb9b2af9e97bb1
Author: Bryan Cutler <[email protected]>
Date: 2017-09-29T15:31:33Z
ARROW-1619: [Java] Set lastSet in JsonFileReader
When reading a vector in JsonFileReader, lastSet should be set in
VariableWidthVectors after reading inner vectors or else subsequent operations
could corrupt the offsets. This also allows to simplify some of the related
code. Additionally, ListVector.lastSet should be explicitly initialized to 0,
which is it's starting offset.
Author: Bryan Cutler <[email protected]>
Closes #1140 from BryanCutler/java-JsonReader-setLast-ARROW-1619 and
squashes the following commits:
8f97a3db [Bryan Cutler] added test for VarBinaryVector that checks lastSet
after reading
70df0cc4 [Bryan Cutler] set lastSet in JsonFileReader and initialize
lastSet for ListVector
commit d4e09c7654030f459f57d330bec639699eadcef6
Author: Rene Sugar <[email protected]>
Date: 2017-09-29T20:29:38Z
ARROW-1615 Added BUILD_WARNING_LEVEL and BUILD_WARNING_FLAGS to Setup…
…CxxFlags.cmake
Author: Rene Sugar <[email protected]>
Closes #1145 from renesugar/ARROW-1615 and squashes the following commits:
71a615e3 [Rene Sugar] ARROW-1615 Add -Wno-vla-extension and change
non-checkin builds back to -Wall
18958430 [Rene Sugar] ARROW-1615 Add -Wno-cast-align
5fe4e8e8 [Rene Sugar] ARROW-1615 Move -Wno-shorten-64-to-32 after
-Wconversion
9d3c7ec3 [Rene Sugar] ARROW-1615 Identify compiler version for clang-802
plus more warning entries
5ebaf86e [Rene Sugar] ARROW-1615 Moved version specific warning entry
971e61aa [Rene Sugar] ARROW-1615 Fixed version specific warning entry
6cf24977 [Rene Sugar] ARROW-1615 Added more version specific Clang warning
entries
50def439 [Rene Sugar] ARROW-1615 Updated build warning level terminology
ea906eb4 [Rene Sugar] ARROW-1615 Check compiler version before disabling
some warnings
159e1897 [Rene Sugar] ARROW-1615 Include CompilerInfo before SetupCxxFlags
in arrow/python
8359c966 [Rene Sugar] ARROW-1615 Added BUILD_WARNING_LEVEL and
BUILD_WARNING_FLAGS to SetupCxxFlags.cmake
commit 7c616114fb83d02faa5921db0296ca994a9b232b
Author: Wes McKinney <[email protected]>
Date: 2017-09-30T04:01:42Z
ARROW-1600: [C++] Add Buffer constructor that wraps std::string
Many other libraries interchange binary data with `std::string`. This makes
it easy to wrap such data in an `arrow::Buffer`.
It may be worth adding a function that creates a buffer from a string, but
owns its memory.
I also deprecated `arrow::GetBufferFromString`, which shouldn't have been
public in the first place, since the new ctor is more general
Author: Wes McKinney <[email protected]>
Closes #1147 from wesm/ARROW-1600 and squashes the following commits:
f60f502c [Wes McKinney] Remove TestBuffer fixture
644bf2b7 [Wes McKinney] Add Buffer ctor that wraps std::string
commit 796129b4f0f714fdb3c4fbf5bc2d2deb55424a84
Author: Wes McKinney <[email protected]>
Date: 2017-09-30T04:02:58Z
ARROW-838: [Python] Expand pyarrow.array to handle NumPy arrays not
originating in pandas
This unifies the ingest path for 1D data into `pyarrow.array`. I added the
argument `from_pandas` to turn null sentinel checking on or off:
```
In [8]: arr = np.random.randn(10000000)
In [9]: arr[::3] = np.nan
In [10]: arr2 = pa.array(arr)
In [11]: arr2.null_count
Out[11]: 0
In [12]: %timeit arr2 = pa.array(arr)
The slowest run took 5.43 times longer than the fastest. This could mean
that an intermediate result is being cached.
10000 loops, best of 3: 68.4 µs per loop
In [13]: arr2 = pa.array(arr, from_pandas=True)
In [14]: arr2.null_count
Out[14]: 3333334
In [15]: %timeit arr2 = pa.array(arr, from_pandas=True)
1 loop, best of 3: 228 ms per loop
```
When the data is contiguous, it is always zero-copy, but then
`from_pandas=True` and no null mask is passed, then a null bitmap is
constructed and populated.
This also permits sequence reads into integers smaller than int64:
```
In [17]: pa.array([1, 2, 3, 4], type='i1')
Out[17]:
<pyarrow.lib.Int8Array object at 0x7ffa1c1c65e8>
[
1,
2,
3,
4
]
```
Oh, I also added NumPy-like string type aliases:
```
In [18]: pa.int32() == 'i4'
Out[18]: True
```
Author: Wes McKinney <[email protected]>
Closes #1146 from wesm/expand-py-array-method and squashes the following
commits:
1570e525 [Wes McKinney] Code review comments
d3bbb3c3 [Wes McKinney] Handle type aliases in cast, too
797f0151 [Wes McKinney] Allow null checking to be skipped with
from_pandas=False in pyarrow.array
f2802fc7 [Wes McKinney] Cleaner codepath for numpy->arrow conversions
587c575a [Wes McKinney] Add direct types sequence converters for more data
types
cf40b767 [Wes McKinney] Add type aliases, some unit tests
7b530e4b [Wes McKinney] Consolidate both sequence and ndarray/Series/Index
conversion in pyarrow.Array
commit 4e0f799e72048587715cb50abdd8d239f3d46d13
Author: Rene Sugar <[email protected]>
Date: 2017-09-30T16:42:37Z
ARROW-1626 Add make targets to run the inter-procedural static analys…
…is tool called infer
Author: Rene Sugar <[email protected]>
Closes #1149 from renesugar/infer and squashes the following commits:
8591b5ff [Rene Sugar] ARROW-1626 Add make targets to run the
inter-procedural static analysis tool called infer
commit 84e5e02fbf412c979387b0a53b0ad0c6d5c5e790
Author: Wes McKinney <[email protected]>
Date: 2017-09-30T18:25:04Z
ARROW-1624: [C++] Fix build on LLVM 4.0, remove some clang warning
suppressions
I'm going to quick make a pass through later today and see how many of
these warning suppressions I can remove
Author: Wes McKinney <[email protected]>
Closes #1148 from wesm/warning-fixes and squashes the following commits:
e930152a [Wes McKinney] Only build compute modules if -DARROW_COMPUTE=ON
d6ca7ac0 [Wes McKinney] Slight refactor of CMakeLists.txt to move Arrow
library setup to src/arrow
e2c61a2e [Wes McKinney] Use -Wno-unknown-warning-option
3d2d7265 [Wes McKinney] Fix travis CI script
6d0d4117 [Wes McKinney] Use BUILD_WARNING_LEVEL in Travis CI
1bec4a76 [Wes McKinney] Fix some more compiler warnings
cae05fb5 [Wes McKinney] Fix documentation compiler warnings
f76b2b93 [Wes McKinney] Fix a bunch of documentation warnings
ac54e2df [Wes McKinney] Remove some clang warning suppressions, fix warnings
6935b8c8 [Wes McKinney] Fix compiler warnings with clang-4.0
commit 0c8b861f93884f2868eb631d8fceee3a8b8905ec
Author: Wes McKinney <[email protected]>
Date: 2017-10-02T00:32:53Z
ARROW-1629: [C++] Add miscellaneous DCHECKs and minor changes based on
infer tool output
This was an interesting journey through some esoterica. I went through all
the warnings/errors that infer (fbinfer.com) outputs and made changes if it
seemed warranted. Some of the checks might be overkill.
See https://gist.github.com/wesm/fc6809e4f4aaef3ecfeb21b8123627bc for a
summary of actions on each warning
Most of the errors that Infer wasn't happy about were already addressed by
DCHECKs. This was useful to go through all these cases -- in nearly all cases
the null references are impossible or would be the result of an error on behalf
of the application programmer. For example: we do not do array boundschecking
in most cases in production builds, but these boundschecks are included in
debug builds to assist with catching bugs caused by improper use by application
developers.
As a matter of convention, we should not use error `Status` to do parameter
validation or asserting pre-conditions that are the responsibility of the
library user. If parameter validation is required in binding code (e.g.
Python), then this validation should happen in the binding layer, not in the
core C++ library.
There are some other cases where we have a `std::shared_ptr<T>` out
variable with code like:
```
RETURN_NOT_OK(Foo(..., &out));
out->Method(...);
```
Here, infer complains that `out` could contain a null pointer, but our
contract with developers is that if `Foo` returns successfully that `out` is
non-null.
Interestingly, infer doesn't like some stack variables that are bound in
C++11 lambda expressions. I noted these in the gist with `LAMBDA`.
Author: Wes McKinney <[email protected]>
Closes #1151 from wesm/fix-infer-issues and squashes the following commits:
f285be95 [Wes McKinney] Restore code paths for empty chunked arrays for
backwards compat
5aa86ce2 [Wes McKinney] More DCHECK esoterica / tweaks based on infer report
22c5d361 [Wes McKinney] Address a couple more infer warnings
75131a6b [Wes McKinney] Some more minor infer fixes
5ff3e3a5 [Wes McKinney] Compilation fix
05316ce4 [Wes McKinney] Fix miscellaneous things that infer does not like.
Make some Python helper functions internal / non-exported
commit a29b0db252f439ff472600c9332006266f601591
Author: Ofek Lev <[email protected]>
Date: 2017-09-26T17:11:49Z
[Python] Update README.md to reflect that wheels are available on all
platforms
Close #1136
Change-Id: Ib9261972fb720df351931902a1666f23b0a0132f
commit 49e02d27227332b06528816bbf73e434a4e1ebcb
Author: Philipp Moritz <[email protected]>
Date: 2017-10-02T12:09:47Z
ARROW-1625: [Serialization] Support OrderedDict and defaultdict
serialization
This PR adds support for OrderedDicts and default dicts using custom
serialization handlers.
Author: Philipp Moritz <[email protected]>
Closes #1152 from pcmoritz/pydict-exact2 and squashes the following commits:
431e0272 [Philipp Moritz] make cloudpickle optional
052b1aa9 [Philipp Moritz] I'd prefer this not to be a runtime dependency
db19ab9b [Philipp Moritz] add tests
799d983e [Philipp Moritz] do not interpret OrderedDict as dict
commit cf738f57e408333db960e0632524437eb7c62d7f
Author: siddharth <[email protected]>
Date: 2017-10-04T17:46:06Z
ARROW-1473: initial prototype hierarchy
----
> [JAVA] Create Prototype Code Hierarchy (alt A)
> ----------------------------------------------
>
> Key: ARROW-1473
> URL: https://issues.apache.org/jira/browse/ARROW-1473
> Project: Apache Arrow
> Issue Type: Sub-task
> Reporter: Jacques Nadeau
> Labels: pull-request-available
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
