You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I've seen some discussion about storage of analysis results and events and so on.
I just realize, that this is something, which we might get from ctapipe for free, if we use the containers correctly.
By correctly I mean the opposite of what I just recently said in a different discussion. So you see, I am often totally lost. :-(
Anyway, I think we agree. We are slowly running into the situation where we have too much custom code for storing data, which leads to files, which can only be read, when closely reading the code, which wrote the files... I think we agree, we would like to avoid that.
At the moment I see two different types of output:
event-wise
cumulative.
For the event-wise output I think we can profit from ctapipe. I'll review their current status and try to propose something. Of course, one often does not want to store the entire event, so a useful solution needs to be able to provide some description like: "store everyting, but not the adc_samples" ... or so.
For the "cumulative" output an example would be the output generated by io.save_bias_curve or io.save_adc. And I have at the moment no idea how to generalize this. So I'd say for this part we just keep on doing whatever we like, and see how it goes. (I need to prevent myself constantly from "overdesigning" ... )
The text was updated successfully, but these errors were encountered:
I've seen some discussion about storage of analysis results and events and so on.
I just realize, that this is something, which we might get from ctapipe for free, if we use the containers correctly.
By correctly I mean the opposite of what I just recently said in a different discussion. So you see, I am often totally lost. :-(
Anyway, I think we agree. We are slowly running into the situation where we have too much custom code for storing data, which leads to files, which can only be read, when closely reading the code, which wrote the files... I think we agree, we would like to avoid that.
At the moment I see two different types of output:
For the event-wise output I think we can profit from ctapipe. I'll review their current status and try to propose something. Of course, one often does not want to store the entire event, so a useful solution needs to be able to provide some description like: "store everyting, but not the adc_samples" ... or so.
For the "cumulative" output an example would be the output generated by
io.save_bias_curve
orio.save_adc
. And I have at the moment no idea how to generalize this. So I'd say for this part we just keep on doing whatever we like, and see how it goes. (I need to prevent myself constantly from "overdesigning" ... )The text was updated successfully, but these errors were encountered: