Duplicate rows when joining two joined streams

[AntonOellerer]

Describe the bug

I wrote a pipeline which reads data from 4 ports, joins two streams at a time into one stream, and then joins the two resulting streams again. When fetching the resulting rows, each row is read up to four times.

To Reproduce

1. Checkout https://github.com/AntonOellerer/SpringQL/tree/bug/duplicate_rows_when_joining_joined
2. Run cargo run --example test_join

Expected behavior

Each timestep one row with the results from the 3 joins is read

Actual behavior

Each timestep up to 4 rows with the results from the 3 joins are read (often also with one column missing)

e.g.:

[motor_average_readings]        2022-01-01 13:00:00.000000000   5       -1      15      19
[motor_average_readings]        2022-01-01 13:00:00.000000000   5       -1      15      19
[motor_average_readings]        2022-01-01 13:00:10.000000000   50      -1      70      -1
[motor_average_readings]        2022-01-01 13:00:10.000000000   50      -1      70      -1
[motor_average_readings]        2022-01-01 13:00:10.000000000   50      -1      70      -1
[motor_average_readings]        2022-01-01 13:00:10.000000000   50      -1      70      -1
[motor_average_readings]        2022-01-01 13:00:30.000000000   15      36      17      -1
[motor_average_readings]        2022-01-01 13:00:30.000000000   15      36      17      -1
[motor_average_readings]        2022-01-01 13:00:30.000000000   15      36      17      -1
[motor_average_readings]        2022-01-01 13:00:30.000000000   15      36      17      -1

Environment

SpringQL

• springql-core: 0.18.1

your machine

• Device: Tuxedo Pulse 15
• OS: Fedora 37

Additional context

• Kernel: 6.1.15
• rustc: 1.68.0

[AntonOellerer]

There generally seems to be a problem with the output produced by joins.
When looking at the springql-web-console, it seems as if each pump produces far more messages than necessary:

In this example, each source received one message per second for 30 seconds