Meeting Updates from 1/29/2024

[garrettmflynn]

This PR implements the cosmetic updates—as well as a few related bugs—from yesterday's meeting, including:

1. Make table borders darker (to see when both are red)
2. Remove n fields counter
3. Make the accordions less obtrusive (remove padding)
4. Ensure all top-level metadata is required (so no exclusion is possible)
5. Remove control + control_description + comments everywhere
6. Replace checkbox with “Skip” button
7. Fix validation using Species.sex on the Subject Table as a test case
8. Ensure you can submit global metadata with loose requirements (i.e. not validated when empty)

Remaining issues that can be tackled in a follow-up include:

1. Provide a better description for what resolution means
2. Ensure that DfOverF—as well as the other nested Ophys properties—cannot be skipped (required, if present, on the NeuroConv side)
3. Provide a test case for multiple imaging planes

[CodyCBakerPhD]

Make the accordions less obtrusive (remove padding)

Still looks like it could go down by another increment on this

[CodyCBakerPhD]

Also tests failing; does that represent a back-compatibility break?

[garrettmflynn]

represent

Since the proper validation has been restored, the tests were failing because ElectrodeColumns was not specified on the test data but now we're correctly trying to check this to see if all our columns are valid.

I've just ignored the check if ElectrodeColumns is blank.

[garrettmflynn]

Good here? Turns out I'd been holding it at a min-height value which kept some "padding" despite removing any explicit declaration.

[garrettmflynn]

The height is still able to grow to fit auxiliary elements like the Skip button.

[CodyCBakerPhD]

Good here? Turns out I'd been holding it at a min-height value which kept some "padding" despite removing any explicit declaration.

Better, maybe one more increment?

Also is it possible to make that box a bit darker gray than the background?

Also note that for DfOverF/Fluorescence/SegmentationImages they weren't supposed to be 'skippable'; that change was only for the source data selector

[garrettmflynn]

Better, maybe one more increment?
I can remove the padding entirely, but this will make things difficult to read and click. The previous image was at 5px at the top and bottom.

Here is 0px.

This is 3px.

Also note that for DfOverF/Fluorescence/SegmentationImages they weren't supposed to be 'skippable'; that change was only for the source data selector
I thought we'd agreed that this is best dealt with through updates to the way that NeuroConv handles the schema. I can patch over this on the GUIDE end, but I didn't want to forget that we have some issues with the schema at a lower level.

[CodyCBakerPhD]

I thought we'd agreed that this is best dealt with through updates to the way that NeuroConv handles the schema. I can patch over this on the GUIDE end, but I didn't want to forget that we have some issues with the schema at a lower level.

What specific issue on the NeuroConv side? If the NeuroConv schema request returns it we can proceed to assume it's required for the GUIDE

[garrettmflynn]

Is this dark enough? Happy to take an rgb value from you, this is rgb(235, 235, 235)

[garrettmflynn]

What specific issue on the NeuroConv side? If the NeuroConv schema request returns it we can proceed to assume it's required for the GUIDE

As Ben mentioned yesterday, if there's no way to not pass DfOverF or any of the other properties (e.g. Subject, Ophys, etc., both of which I've patched because they're signficantly more noticeable) to NeuroConv, the schema should include it in the required list.

This would require dynamically changing the returned schema, which may present a more substantial change that we can readily implement now.

[garrettmflynn]

Alright just pushed an update to require all sub-levels of the Ophys metadata, if included in the schema.

[garrettmflynn]

Also once the accordion is opened, it doesn't matter how small the padding was because the content pushes all the others beneath the fold.

IMO the size of these dropdown headers now aren't well-distinguished from the content, making it all more difficult to process once open. Though these screenshots don't really do this justice.

[CodyCBakerPhD]

Is this dark enough? Happy to take an rgb value from you, this is rgb(235, 235, 235)

This is pretty bizarre... Maybe it's happening with your screenshots too, but when I take a screenshot of what it looks like to me the gray gets whitened

(the one that looks gray there is from me hovering over it, which was even a shade darker than the default grayness)

On my screen, the current gray looks OK

3px on the padding looks good

Alright just pushed an update to require all sub-levels of the Ophys metadata, if included in the schema.

Thanks; it will take quite some thought about how to do something similar on the NeuroConv side since the way you interact with metadata is starting to become very very different. For example, if the key reference to a plane changes in the metadata instance (which is perfectly allowed in the programmatic side) you'd also have to update the requirements to exactly match that update to the pattern property in the schema I think. Which would be really weird

In the end I think they are becoming two separate things: the GUIDE is a very limited scoped way of altering metadata fields on a fixed instance of the metadata dict, constrained by the default schema for each interface. So I think saying the required = properties here specific to the GUIDE is perfectly OK

[CodyCBakerPhD]

IMO the size of these dropdown headers now aren't well-distinguished from the content, making it all more difficult to process once open.

Fixable by making the font size of the headers slightly larger, no?

[garrettmflynn]

This is pretty bizarre... Maybe it's happening with your screenshots too, but when I take a screenshot of what it looks like to me the gray gets whitened

Hmmm weird. I used a color picker on your image and it's still the original rgb(240,240,240). Must be a perceptual illusion as a result of all the other grays in the application.

[garrettmflynn]

Fixable by making the font size of the headers slightly larger, no?

Actually, it looks best to me if I reset the location of the colored status bar to the bottom (rather than the left). Changing the font size will start messing with the visual balance, requiring more padding to look intentionally designed.

[CodyCBakerPhD]

Actually, it looks best to me if I reset the location of the colored status bar to the bottom (rather than the left). Changing the font size will start messing with the visual balance, requiring more padding to look intentionally designed.

Oky doky

Just keep in mind I'm still of the opinion that top level tabs (like the way the pydata docs theme looks) instead of accordions would remedy all of these issues

[garrettmflynn]

Right. I'll add that back to my to-dos and see if I can get that implemented shortly.

[CodyCBakerPhD]

Right. I'll add that back to my to-dos and see if I can get that implemented shortly.

Sounds good, no rush if it's tricky. Was just a dream

The rest of these cosmetic updates look/feel good