search

fermi-ad / acsys-python

Python module to access the Fermilab Control System
MIT License
8 stars 4 forks source link

What is ItemData.cycle represent? #39

Closed kjhazelwood closed 2 years ago

kjhazelwood commented 2 years ago

I see in the README.md for acsys.dpm a attribute of ItemData called cycle. I dont see it available in 0.11.4, I presume its not available until v1.0? What does cycle represent? Iteration count of the dpm.replies() genenerator? Or an actual beam cycle id (that would be even better, but not sure what that would be from)?

rneswold commented 2 years ago

It was to be the 15 Hz Booster cycle number.

In v1.x, we removed this field because ItemData now always has an array of (stamp, value) pairs. For slow rates, the array will have one element. Faster than 15 Hz will switch to the FTP protocol and ItemData's array will have more than one element.

The cycle number doesn't easily fit in this model (since you could ask for 100Hz data which can span cycles.)

Maybe each ItemData can have a single cycle value. ~For <= 15hz, it'll be the cycle associated with the single piece of data. For faster rates, It's the cycle for the last point and you can work back through the data if you care about the cycle values for each point (but for FTP data, I doubt anyone wants to correlate the cycle number.)~ The cycle value is always associated with the last item in the array.

rneswold commented 2 years ago

I built a test version in which ItemData gets the .cycle value.

For slow rates, it appears to be a millisecond timestamp from the Epoch. For fast rates, it's just a packet counter. I don't mind committing this code, but maybe we can convince @charlieking65 to make DPM return the actual cycle number of the data.

kjhazelwood commented 2 years ago

Thanks for the explanation. For my purposes, even at FTP rates, a 15Hz cycle_id would be very beneficial. Data synchronization even at high data rates usually only has to be attributed to the same Booster injection. I think FTP itself does a pretty good job for synchronizing data on the mS level.

rneswold commented 2 years ago

We need to get the DPM project onto GitHub so we can transfer this issue to it. 😄

beauremus commented 2 years ago

I think it's fine to leave this here, but I created a DPM repo here if you want to transfer this issue https://github.com/fermi-controls/data-pool-manager.

rneswold commented 2 years ago

aeca61c496739d14a585f5c1626d93d4b0e8abf3 makes the cycle value available in ItemData's .cycle property. The rest of this issue will be handled in the issue mentioned above.