ISO8601 time string with a Z comes out with +00:00 after load and then dump

When using the fields.DateTime and a value "2018-07-25T00:00:00Z". Loading and then dumping that data should result in the exact same value for that field. But it doesn’t. What is dumped back out is "2018-07-25T00:00:00+00:00". This is not technically a different time, but I think I should be able to serialize/deserialize data and not have it be altered from the voyage.

datetime strings are really difficult with python. My sympathies.

I’m going to try to make a subclass of field that does only what I want for tz-aware iso8601 strings and isn’t as broad as fields.DateTime. I’ll post it back for reference.

[deckar01]

Loading and then dumping that data should result in the exact same value for that field. But it doesn’t.

It is the same value with different encodings. Z and +00:00 are equivalent.

import dateutil

dt = dateutil.parser.parse('2018-07-25T00:00:00Z')
dt.tzinfo()
# tzutc()
dt.isoformat()
# 2018-07-25T00:00:00+00:00

dateutil’s parser is passing the timezone correctly and datetime.isoformat() is allowed to represent Z as +00:00. If you are interfacing with a system that doesn’t accept +00:00 as a timezone, it probably isn’t compliant. Otherwise it is a personal preference that can be achieved with a custom format.

[sloria]

Given the current design, there isn’t a good way outside of passing format to DateTime that ensures that the input and output format are exactly the same. If you need more complex logic that format, a custom field is the way to go.