This is an automated email from the ASF dual-hosted git repository.
yangzhg pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-doris.git
The following commit(s) were added to refs/heads/master by this push:
new 2010d33 [Doc] Translate a Chinese statement which appears in English
version doc (#5290)
2010d33 is described below
commit 2010d331d36d027b76afdac064d447aa29502682
Author: Ting Sun <[email protected]>
AuthorDate: Tue Jan 26 09:13:30 2021 +0800
[Doc] Translate a Chinese statement which appears in English version doc
(#5290)
---
docs/en/internal/doris_storage_optimization.md | 4 ++--
docs/en/internal/grouping_sets_design.md | 10 +++++-----
docs/en/internal/metadata-design.md | 2 +-
3 files changed, 8 insertions(+), 8 deletions(-)
diff --git a/docs/en/internal/doris_storage_optimization.md
b/docs/en/internal/doris_storage_optimization.md
index 1dbf697..4cfb586 100644
--- a/docs/en/internal/doris_storage_optimization.md
+++ b/docs/en/internal/doris_storage_optimization.md
@@ -216,7 +216,7 @@ Relevant issues:
Relevant issues:
1. How to quickly locate a row within the page?
- The data inside the page is encoding, so it can not locate the
row-level data quickly. Different encoding methods have different schemes for
fast line number positioning in-house, which need to be analyzed concretely:
+ The data inside the page is encoding, so it cannot locate the row-level
data quickly. Different encoding methods have different schemes for fast line
number positioning in-house, which need to be analyzed concretely:
- If it is rle-coded, skip is performed by resolving the head of RLE
until the RLE block containing the row is reached, and then the reverse
solution is performed.
- binary plain encoding: offset information will be stored in the page,
and offset information will be specified in the page header. When reading,
offset information will be parsed into the array first, so that you can quickly
locate the data of a row of block through offset data information of each row.
2. How to achieve efficient block reading? Consider merging adjacent blocks
while they are being read, one-time reading?
@@ -232,4 +232,4 @@ It implements a scalable compression framework, supports a
variety of compressio
## TODO ##
1. How to implement nested types? How to locate line numbers in nested types?
-2. How to optimize the downstream bitmap and column statistics statistics
caused by ScanRange splitting?
+2. How to optimize the downstream bitmap and column statistics caused by
ScanRange splitting?
diff --git a/docs/en/internal/grouping_sets_design.md
b/docs/en/internal/grouping_sets_design.md
index bbbe74b..f1a3b8b 100644
--- a/docs/en/internal/grouping_sets_design.md
+++ b/docs/en/internal/grouping_sets_design.md
@@ -261,7 +261,7 @@ Presto supports composition, but not nesting.
## 2. Object
Support `GROUPING SETS`， `ROLLUP` and `CUBE ` syntax，implements 1.1, 1.2, 1.3
1.4, 1.5, not support the combination
- and nesting of GROUPING SETS at current version.
+ and nesting of GROUPING SETS in current version.
### 2.1 GROUPING SETS Syntax
@@ -308,7 +308,7 @@ Expression，column name.
### 3.1 Overall Design Approaches
-For `GROUPING SET` is equivalent to the `UNION` of `GROUP BY` . So we can
expand input rows, and run an GROUP BY on these rows。
+For `GROUPING SET` is equivalent to the `UNION` of `GROUP BY` . So we can
expand input rows, and run an GROUP BY on these rows.
For example：
@@ -316,7 +316,7 @@ For example：
SELECT a, b FROM src GROUP BY a, b GROUPING SETS ((a, b), (a), (b), ());
```
-Data in table src :
+Data in table src :
```
1, 2
@@ -485,7 +485,7 @@ select NULL, NULL, sum(k3) from t;
In order to add GROUPING_ID to groupingExprs in GroupByClause, need to create
virtual SlotRef, also, need tot create a tuple for this slot, named
GROUPING\_\_ID Tuple.
-For the plannode RepeatNode, it's input is all the tuple of it's children,
It's output tuple is the repeat data and GROUPING_ID.
+For the plannode RepeatNode, its input are all the tuples of its children and
its output tuple are the repeat data and GROUPING_ID.
#### 3.3.3 Expression and Function Substitution
@@ -498,4 +498,4 @@ grouping_id() -> grouping_id(grouping_id) for grouping_id
function
#### 3.4.1 Tasks
1. Add RepeatNode executor, expend the input data and append GROUPING_ID to
every row
-2. Implements grouping_id() and grouping() function.
\ No newline at end of file
+2. Implements grouping_id() and grouping() function.
diff --git a/docs/en/internal/metadata-design.md
b/docs/en/internal/metadata-design.md
index d02d5d3..459cd74 100644
--- a/docs/en/internal/metadata-design.md
+++ b/docs/en/internal/metadata-design.md
@@ -38,7 +38,7 @@ under the License.
As shown above, Doris's overall architecture is divided into two layers.
Multiple FEs form the first tier, providing lateral expansion and high
availability of FE. Multiple BEs form the second layer, which is responsible
for data storage and management. This paper mainly introduces the design and
implementation of metadata in FE layer.
-1. FE 节点分为 follower 和 observer 两类。各个 FE 之间，通过 bdbje（[BerkeleyDB Java
Edition](http://www.oracle.com/technetwork/database/database-technologies/berkeleydb/overview/index-093405.html)）进行
leader 选举，数据同步等工作。
+1. There are two different kinds of FE nodes: follower and observer. Leader
election and data synchronization are taken among FE nodes by bdbje
([BerkeleyDB Java
Edition](http://www.oracle.com/technetwork/database/database-technologies/berkeleydb/overview/index-093405.html)).
2. The follower node is elected, and one of the followers becomes the leader
node, which is responsible for the writing of metadata. When the leader node
goes down, other follower nodes re-elect a leader to ensure high availability
of services.
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
