add journal table and JournalWriter helper class #89
Conversation
| if event_type is None: | ||
| if payload is None: | ||
| event_type = "null" | ||
| else: | ||
| event_type = type(payload).__name__ | ||
|
|
||
| if payload.person_id is None: | ||
| payload.person_id = self.person_id | ||
| if payload.post_id is None: | ||
| payload.post_id = post_id | ||
| if payload.workpackage_id is None: | ||
| payload.workpackage_id = workpackage_id | ||
| if payload.task_type is None: | ||
| payload.task_type = task_type |
There was a problem hiding this comment.
You can write it in a shorter way :
if event_type is None:
event_type = "null" if payload is None else type(payload).__name__
payload.person_id = payload.person_id or self.person_id
payload.post_id = payload.post_id or post_id
payload.workpackage_id = payload.workpackage_id or workpackage_id
payload.task_type = payload.task_type or task_typeThere was a problem hiding this comment.
Thank for the tip. Just to keep it short I could even also write
event_type = payload and type(payload).__name__ or "null"
It is a matter of taste ... I am personally not the biggest fan of the python ternary operator.
| def generate_time_uuid(node=None, clock_seq=None): | ||
| """Create a lexicographically sortable time ordered custom (non-standard) UUID by reordering the timestamp fields of a version 1 UUID.""" | ||
| (time_low, time_mid, time_hi_version, clock_seq_hi_variant, clock_seq_low, node) = uuid1(node, clock_seq).fields | ||
| # reconstruct 60 bit timestamp, see version 1 uuid: https://www.rfc-editor.org/rfc/rfc4122 | ||
| timestamp = (time_hi_version & 0xFFF) << 48 | (time_mid << 32) | time_low | ||
| version = time_hi_version >> 12 | ||
| assert version == 1 | ||
| a = timestamp >> 28 # bits 28-59 | ||
| b = (timestamp >> 12) & 0xFFFF # bits 12-27 | ||
| c = timestamp & 0xFFF # bits 0-11 (clear version bits) | ||
| clock_seq_hi_variant &= 0xF # (clear variant bits) | ||
| return UUID(fields=(a, b, c, clock_seq_hi_variant, clock_seq_low, node), version=None) |
There was a problem hiding this comment.
What is the point of using time ordered uuid? Shouldn't we rely on created_date if we ever need to do something related to the time (e.g. weekly leaderboard).
There was a problem hiding this comment.
You may know the snowflake datatype (as used by twitter or discord). The idea here is the same: Time ordered unique ids that can be client-side generated on different machines. It will be used by integrator processes to track up to which point in time events have been processed.
Log user responses chronologically in a journal table #72. Ratings and Rankings are directly stored in the journal, text-replies currently not.