Make peaklets dtype flexiable #1299
Conversation
|
Edit: We still need the if statement in merged_s2s because the functions which build the merged S2s require the field... |
|
I am happy to review this one. Isn't this PR more about |
|
@FaroutYLq thanks, yes it only changes the lineage of everything above peaklets. I also have an idea how to fix the remaining test failure. This requires a change in strax though. I hope I can add it today. |
|
@FaroutYLq this PR requires: AxFoundation/strax#785 once it is merged it should work. If you have time it would be nice if you could review the other one as well. |
…raxen into make_peaklets_dtype_flexiable
|
I changed back the plugin version and added back in redundant option to preserve the lineage. |
| sum_waveform_top_array = straxen.URLConfig( | ||
| default=True, type=bool, help="Digitize the sum waveform of the top array separately" | ||
| ) | ||
|
|
||
| n_top_pmts = straxen.URLConfig(type=int, help="Number of top TPC array PMTs") |
There was a problem hiding this comment.
Why these lines are moved several lines below?
There was a problem hiding this comment.
Ah because I added again back in sum_waveform_top_array must have added it at the wrong position. Otherwise the lineage would change. Even though the option is not needed anymore here.
There was a problem hiding this comment.
I would approve the PR if moved back. :)
There was a problem hiding this comment.
Very picky, I like it ;) :D I will do just a sec.
|
I have checked that the lineage of |
Side note this PR will change the lineage of everything > peaklets!
What does the code in this PR do / what does it improve?
Simple PR which makes the dtype propagation for peaklets to peaks more flexible. In the current implementation the dtype of merged S2s and peaks is fixed to the
peaklets_dtypewhich is fixed to thepeaks_dtypedefined in strax. Already in the past this lead to an issue when we added new optional fields to peaks. For example when adding the hit timing https://github.com/AxFoundation/strax/blame/21cc96e011b0e4099138979791f34e8b1addedb7/strax/dtypes.py#L211 or top waveform https://github.com/AxFoundation/strax/blame/21cc96e011b0e4099138979791f34e8b1addedb7/strax/dtypes.py#L222. Especially, for the latter we added an if statement intomerged_S2sto also include thetop_datafield when doing so on the peaklet level.Also for the SOM peaklet_classifcation this lead to issues because no additional fields could be added to
peaklet_classifcationwithout also changing thepeakletdata-type.This is in general a not very strax-like behavior in which we know for each plugin its dependencies and the data-types of these dependencies.
Can you briefly describe how it works?
To circumvent this issue and to make the peaklet and peaklet_classification data-type more flexiable, I compute now the merged_S2s dtype based on the peaklets and peaklet_classifcation dtypes of the currently registered plugins. This also allows us to remove the additional if-statement in merged_s2s needed for the top waveform. In this way analysts can add as many fields to their own peaklet and classification plugins as they like. To be still able to control which parameters are propagated from peaklets to peaks (and higher up) I added a copy statement at the end of peaks which only copies all fields from peaklets which are provided by the Peaks plugin. The additional computation time is negligible.
Although the changes do not change the data provided by the plugins, I bumped the code version.