Allow unit vectordata #506
Conversation
- renamed variables to for clarity - retabbed function files to include one scope for base/private functions.
If the object's data is not a reference but has region/object references in its attributes, then this error may occur because all reference corrections require calling `export` again. Since we want to limit the possibility of object reference cycles, it's better to write any dataset given that it's not reference data and just check for duplicates before doing so.
Used to error, now short-circuits the output.
Removed the data-class setter format and use the correct handle-class one (without returning self).
For hard-coding VectorData.unit. Also retabbed to maintain consistent scope for functions.
Now that we have hidden properties in our generated classes.
lawrence-mbf
added
category: bug
|
Will you add the same workaround for the https://github.com/NeurodataWithoutBorders/nwb-schema/blob/dev/core/nwb.misc.yaml#L248 |
I can do that but could either @rly or @bendichter draft a python script that creates a Unit table with these attributes so I can add it to the python tests? |
Originally applied to all export calls. This may cause issues for datapipes and simply rewriting data so the check is now applied to datastubs instead.
Codecov Report
@@ Coverage Diff @@
## master #506 +/- ##
==========================================
+ Coverage 87.58% 87.89% +0.30%
==========================================
Files 129 129
Lines 5340 5451 +111
==========================================
+ Hits 4677 4791 +114
+ Misses 663 660 -3
... and 2 files with indirect coverage changes 📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
|
I did a simple spot check on this dataset. Read was broken before and with this change read is fixed. I'll try a few more spot checks. |
|
@yarikoptic would it be possible to run the healthstatus checks against this branch? It appears to work but I want to be thorough |
|
not "easily". We have dandi/dandisets-healthstatus#20 but not yet pursued that direction. |
|
Ueli's lab has confirmed that this works for them, fixing a write command that uses waveform_mean.sampling_rate and waveform_mean.unit |
Currently throws a strange error when testing out to pynwb
Mostly description and default columns were missing. These have been filled in. The getContainer() function was also changed so as to test the entire Units table and not just one column.
Fixes #238
Motivation
See #238 and friends
How to test the behavior?
See #238 and friends. A good example file is DANDIset 000022
waveform_mean,waveform_sd, andwaveformsattributes:unitsampling_ratespike_timesattributes:resolutionChecklist
fix #XXwhereXXis the issue number?