Below is a proposed data format to send UniVerse database changes via file. It uses standard UNIX delimiters for new lines, and standard UniVerse characters to deliniate fields (ASCII code 254 (þ)) and data (ASCII code 253 (ý)).
Example:
src=cssþns=CONTRACTSþts_ms=1550549470126þuid=12345678þop=uþdata=ý_id=1234ýname=Bobýþ
src=cssþns=CONTRACTSþts_ms=1550549470126þuid=12345678þop=uþdata=ý_id=1234ýname=Simonýþ
Technical Notes:
- All lines are ended in ASCII code 10 (LF), including the last line
- The file is ended with an EOF (handled by the file system)
A 'change set' is zero or more 'row events', where each row event describes , which describe a series of row changes to one or more 'files' in UniVerse.
Delimiter: ASCII code 10 (LF)
Metadata and data representing the change that has occured.
Delimiter: ASCII code 254 (þ)
- Source (
src) - indicator that the data is coming from UniVerse (Similar to a Database name) - Namespace (
ns) - location from within the source where data is coming from (UniVerse File name) - Timestamp (
ts_ms) - time when row changed (from system clock) (Milliseconds since Epoch) - Incrementing Id (
uid) - unique, incrementing Id that uniquely represents the row trigger event - Operation (
op) - representation of the type of change to the row (Read, Create, Update, Delete) - Data (
data) - set of key/values representing the data that has changed
src=cssþns=CONTRACTSþts_ms=1550549470126þuid=12345678þop=uþdata=ý_id=1234ýname=Bobýþ
Format: "op={operation_type}"
Examples: op=r, op=c, op=u, op=d
- Read (op=r) - An initial
- Create (op=c) - An insert into the table. A new entity/row has been created.
- Update (op=u) -
- Delete (op=d)
Contains the data that has changed. Should generally contain the row's Primary Key with _id
The key/values should be surrounded, both start and end, with the delimiter.
Format: "ýkey1=value1ýkey2=value2ý"
Condition: key must not contain '='
Delimiter: ASCII code 253 (ý)