[BUG] Invalid schema (bad yaml file)

[dipterix]

Describe your problem in detail.

bids-specification/src/schema/objects/enums.yaml

Line 1139 in fa2b5d8

y:

YAML 1.1 interprets y:... as key true instead of literal y, hence this enum will be interpreted as

{true: {
  value: true,
  display_name: true,
  ...
}}

Describe what you expected.

Suggested change:

x:
  value: x
  display_name: x
  description: |
    The x dimension of the coordinate system.
"y":
  value: "y"
  display_name: "y"
  description: |
    The y dimension of the coordinate system.
z:
  value: z
  display_name: z
  description: |
    The z dimension of the coordinate system.

BIDS specification section

https://bids-specification.readthedocs.io/en/latest/...

[effigies]

It seems to have done the right thing when converted to JSON:

https://bids-specification.readthedocs.io/en/latest/schema.json

We don't generally expect people to parse the YAML separately. What is the context where you're hitting problems?

Anyway, we can definitely accept a patch to use quotes. Would be worth adding a comment so they don't get removed in the future.

[dipterix]

Ah I see. I'm using the json now. Thanks!

Can I ask two questions here?

First question is I found that schema.rules.files.raw.anat.nonparametric is identical to schema.rules.files.deriv.imaging.anat_nonparametric_volumetric, I wonder what's the key (unique ID) for the entity table? Somehow I thought the datatype+suffix consists the unique key for the entity table. Now I'm not sure...

The second question is that I'm trying to implement some basic rules from schema so I don't have to maintain them manually when BIDS version changes. I noticed that in schema, there are the following "contexts": (according to https://bidsschematools.readthedocs.io/en/latest/description.html#on-reused-objects-with-different-definitions <term>_<context> syntax, e.g. timeseries__mri_no_task)

 [1] "scans"         "sessions"      "channels"      "optode"        "entities"      "channel"      
 [7] "electrodes"    "optodes"       "segmentations" "motion"        "eeg"           "ieeg"         
[13] "nirs"          "mtransfer"     "mri"           "coordsystem"   "fmap"          "ds_relative"  
[19] "mrs"           "meg_common"    "motion_common" "eeg_common"    "meg"           "micr"         
[25] "mri_no_task"   "pet"           "anat"          "func"    

I wonder where can I get these context string, or how are they defined?

[effigies]

First question is I found that schema.rules.files.raw.anat.nonparametric is identical to schema.rules.files.deriv.imaging.anat_nonparametric_volumetric, I wonder what's the key (unique ID) for the entity table? Somehow I thought the datatype+suffix consists the unique key for the entity table.

The difference between these is that anat_nonparametric_volumetric should also have space, desc, resolution and density entities, and is valid if dataset_description.DatasetType == "derivative":

bids-specification/src/schema/rules/files/deriv/imaging.yaml

Lines 2 to 9 in 79e4ea8

	anat_parametric_volumetric:
	$ref: rules.files.raw.anat.parametric
	entities:
	$ref: rules.files.raw.anat.parametric.entities
	space: optional
	resolution: optional
	density: optional
	description: optional

There's discussion in #2052 about adding explicit selectors to file rules.

The second question is that I'm trying to implement some basic rules from schema so I don't have to maintain them manually when BIDS version changes. I noticed that in schema, there are the following "contexts": (according to https://bidsschematools.readthedocs.io/en/latest/description.html#on-reused-objects-with-different-definitions <term>_<context> syntax, e.g. timeseries__mri_no_task)

 [1] "scans"         "sessions"      "channels"      "optode"        "entities"      "channel"      
 [7] "electrodes"    "optodes"       "segmentations" "motion"        "eeg"           "ieeg"         
[13] "nirs"          "mtransfer"     "mri"           "coordsystem"   "fmap"          "ds_relative"  
[19] "mrs"           "meg_common"    "motion_common" "eeg_common"    "meg"           "micr"         
[25] "mri_no_task"   "pet"           "anat"          "func"    

They're not defined. They're just mnemonics for giving the terms unique names, which allows them to be referenced elsewhere in the schema.

[dipterix]

Got it. So there are rules for which schema to be used based on context, but they are not yet stale. Thanks for answering my questions promptly, and I will watch/follow up on issue 2052 from now on.

Please feel free to close this comment as all questions have been addressed.