Mateusz Kobos created AVRO-1395:
-----------------------------------
Summary: Order of fields returned by DataFileReader should match
the order defined in Avro schema
Key: AVRO-1395
URL: https://issues.apache.org/jira/browse/AVRO-1395
Project: Avro
Issue Type: Improvement
Components: python
Affects Versions: 1.7.5
Reporter: Mateusz Kobos
Priority: Minor
Python `DataFileReader` class allows to iterate over records of Avro file where
each record is represented by a standard Python dictionary - each element
corresponds to a single field of the record. Note that this dictionary does not
define any particular order of its elements. I claim that it would be better if
the order of elements followed the order of fields as defined in file's Avro
schema. In such case, presentation of the record would be more human-friendly,
which is important e.g. for a user that wants to view the record as a JSON
string. Consider that order of fields in Avro schema usually has some
significance, i.e., the most important fields, (like ID), are at the beginning.
Implementing the functionality of having the representation of the record
follow the order of fields defined in Avro schema seems to be pretty easy,
i.e., as of Avro 1.7.5 in `io.DatumReader` class in `read_record` method, you
would only have to change the second line of code from `read_record = {}` to
`read_record = OrderedDict()`.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
