How to use Artic properly

[octosport]

Arctic Version

Any

Arctic Store

ChunkStore

Platform and version

Linux

Description of problem and/or code sample that reproduces the issue

Although the library seems very powerful it lacks simple production use cases. If I take the example in ChunkStore:

>>> df = DataFrame(data={'data': [1, 2, 3, 4, 5, 6, 7, 8, 9],
                         'open': [1.1, 2.2, 3.3, 4.4, 5.5, 6.6, 7.7, 8.8, 9.9],
                         'close': [1.2, 2.3, 3.4, 4.5, 5.6, 6.7, 7.8, 8.9, 9.0],
                         'prev_close': [.1, .2, .3, .4, .5, .6, .7, .8, .8],
                         'volume': [1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000]
                         },
                   index=MultiIndex.from_tuples([(dt(2016, 1, 1), 1),
                                                 (dt(2016, 1, 2), 1),
                                                 (dt(2016, 1, 3), 1),
                                                 (dt(2016, 2, 1), 1),
                                                 (dt(2016, 2, 2), 1),
                                                 (dt(2016, 2, 3), 1),
                                                 (dt(2016, 3, 1), 1),
                                                 (dt(2016, 3, 2), 1),
                                                 (dt(2016, 3, 3), 1)],
                                                names=['date', 'id'])
                   )

>>> lib.write('column_test', df, chunk_size='D')
>>> lib.read('column_test', columns=['prev_close', 'volume'])

               prev_close  volume
date       id                    
2016-01-01 1          0.1    1000
2016-01-02 1          0.2    2000
2016-01-03 1          0.3    3000
2016-02-01 1          0.4    4000
2016-02-02 1          0.5    5000
2016-02-03 1          0.6    6000
2016-03-01 1          0.7    7000
2016-03-02 1          0.8    8000
2016-03-03 1          0.8    9000

• How do you add a new column, for instance, a high or low, without making a new symbol but updating that one? _It seems possible in ChunkStore using update but we need to precise chunk_range and the new data frame to be written has to contain old columns + the new one. It is not possible to only append one new column without deleting the rest of the data_
• How do you update each column individually on the 2016-03-04 (say I do not get each data point at the same time)? _It seems possible in ChunkStore using update but we need to precise chunk_range otherwise, data are overwritten_
• How do you read multiple symbol volumes and get them in one data frame? It is possible in ChunkStore to read by passing a list of symbols
• How do you update/append in TickStore? And I guess in VersionStore you write a new version for each update? My understanding is TickStore only write as ticks are only read one time

Maybe my understanding is not correct and the database is not designed for these cases which is perfectly fine. Otherwise, I can not find this information in the doc.

Thanks

[vietlq]

@mehertz ^^ probably you need to write one more article to address these requests haha