Implement Issue #389 : Support for MessagePack #460
Conversation
Codecov Report
@@ Coverage Diff @@
## master #460 +/- ##
============================================
+ Coverage 86.63% 86.81% +0.18%
- Complexity 302 309 +7
============================================
Files 20 20
Lines 1294 1327 +33
Branches 135 137 +2
============================================
+ Hits 1121 1152 +31
- Misses 113 115 +2
Partials 60 60
Continue to review full report at Codecov.
|
| @@ -3,6 +3,8 @@ | |||
| import java.util.List; | |||
| import java.util.Map; | |||
|
|
|||
| import com.fasterxml.jackson.annotation.JsonIgnoreProperties; | |||
There was a problem hiding this comment.
Why do we introduce a new dependency to jackson ?
There was a problem hiding this comment.
because the converter https://github.com/komamitsu/retrofit-converter-msgpack (you proppsed it) depends on jackson
There was a problem hiding this comment.
OK, i didnt check that, my fault. But still, using another json converter with this change is not desired. We must think twice "
| import com.fasterxml.jackson.databind.MappingIterator; | ||
| import com.fasterxml.jackson.databind.MappingJsonFactory; | ||
| import com.fasterxml.jackson.databind.ObjectMapper; | ||
| import com.fasterxml.jackson.databind.ObjectReader; |
There was a problem hiding this comment.
Same here, i really would like to prevent to switch from moshi to jackson with this PR. I am not against changing the json converter library, i just want to do it by "accident" when introducing messagepack format. We should do this move, once decided intentionally and without any other modifications.
There was a problem hiding this comment.
again, the convertor depends on Jackson
| result = results.nextValue(); | ||
| consumer.accept(result); | ||
| } | ||
| }; |
There was a problem hiding this comment.
Way too much code in the constructor.
There was a problem hiding this comment.
In a minor release, I would like to keep the implementation as back-ward compatible as possible. this constructor
InfluxDBImpl(final String url, final String username, final String password, final OkHttpClient.Builder client,
final InfluxDBService influxDBService, final JsonAdapter<QueryResult> adapter) {
in my opinion, even it seems awkward and used nowhere in the source code, we are not used if it is used by any user or not so I decided to keep it. That lead me to introduce the ChunkProcessor so we can have polymorphism for processing chunking query
@majst01, I believe Jackson has a better API on processing HTTP chunked response, You can compare 2 implementation of ChunkProccesor to see how different they are
| @@ -130,7 +132,8 @@ public void testFlushDuration() throws InterruptedException { | |||
|
|
|||
| //check no points writen to DB before the flush duration | |||
| QueryResult result = influxDB.query(new Query("select * from weather", dbName)); | |||
| Assertions.assertNull(result.getResults().get(0).getSeries()); | |||
| List<Series> series = result.getResults().get(0).getSeries(); | |||
There was a problem hiding this comment.
Any reason why a test must be changed ?
There was a problem hiding this comment.
Because:
+in JSON response (for an empty query), the 'series' element does not exists
- in MessagePack response (for an empty query), series exist as an empty array
--> I think none of them are not wrong, just the consistency from server side. I am actually thinking to open an issue for the server
@majst01 , I think that's not a fault, because I found that this MessagePack retrofit converter is really documented in Retrofit Wiki https://github.com/square/retrofit/wiki/Converters). So I believe it is somewhat recommended by Retrofit guys. BTW, in a response to your comments. I propose following changes
What do you think ? |
|
closed because PR #471 is the final PR |
thi PR for Issue #389 : Support for MessagePack
I use the https://github.com/komamitsu/retrofit-converter-msgpack (thanks to @majst01)