Set default view configuration in datasource properties json

[youri-k]

URL of deployed dev instance (used for testing):

• https://___.webknossos.xyz

Steps to test:

for the evolutions:

• after the evolution the segmentationOpacity of the configuration element should be gone
• instead the old segmentationOpacity value should be the alpha value of a newly created entry in the layers dict with the name of the related segmentation layer as key
• the reverse evolution should take the alpha value from segmentation layer entry in the layers dict and save it as segmentationOpacity
• to confirm that the configuration entries are still correct, open the dataset edit view and look at the view configurations

for the disk defaults:

• if a view configuration is supplied in the datasource properties json, it should take precedence over the normal default, but if a existing dataset config takes precedence over the disk default
• e.g. change the alpha value of a color layer (default is 100)

Issues:

• fixes Set initial layer color trough datasource-properties.json #4298
• fixes Remove segmentationOpacity from DataSet #4378

---

• Updated changelog
• Updated migration guide if applicable
• [ ] Updated documentation if applicable
• [ ] Adapted wk-connect if datastore API changes
• Needs datastore update after deployment
• Ready for review

[youri-k]

@MichaelBuessemeyer this pr should also take care of #4378. Could you do the necessary frontend part? This involves the the dataset edit view and the rendering in the tracing view. Thank you 🙂

[fm3]

Wow, cool stuff :)
Small thing: could you provide a small example json in the changelog / steps to test?
Also, reading the changelog, I think defaultViewConfigurationOpt is a user-facing field name? I’d prefer to go without the “Opt” in the json (that was just a shorthand for us related to the scala Option code)

[MichaelBuessemeyer]

@youri-k If we want to

Instead use the alpha value of the SegmentationLayer

we need to add the configs of the segmentation layer to the other layer configs in the DatasetConfiguration under the property "layers". Currently, that's not the case and will be needed.

Can you please do the backend part of this and write a conversion script, that removes the "segmentationOpacity" from DatasetConfiguration and adds a layer configuration for the segmentation to the layers property of DatasetConfiguration.

[philippotto]

LGTM! But as @fm3, I didn't test this again. Please do so before deploying (especially the case were view configurations already exist in the old format).

[MichaelBuessemeyer]

@philippotto First of all, I am sorry that I missed merging this PR earlier.

Because I missed this, your newest PR reworking the rendering to only display the enabled layers when the GPU does not support rendering all layers created some "hidden" merge conflicts.

I tried resolving them and it looks fine now and everything works as expected on my local machine. Nevertheless could you please check the newest changes again and whether everything works on your machine? But I can ensure that the bug @fm3 found is resolved and does not appear in the current version of this PR.

[philippotto]

Thanks for resolving the hidden conflicts!

I just tested the branch, but something doesn't seem right to me:

• I edited 2017-05-31_mSEM_scMS109_bk_100um_v01-aniso so that the color layer has alpha: 80 and color: [ 255, 100, 255 ] and made the dataset public.
• When opening https://viewdefaultsfromdisk.webknossos.xyz/datasets/sample_organization/2017-05-31_mSEM_scMS109_bk_100um_v01-aniso/view in an incognito tab, the layer configuration is not the one I edited. Instead, alpha is 100% and color is white.

Shouldn't this use the values I edited?

[MichaelBuessemeyer]

@philippotto Thanks for finding this bug. I debugged it a bit further and it looks like the backend sends false information:

• When I open the link of yours in a private tab and look at the network traffic the https://viewdefaultsfromdisk.webknossos.xyz/api/dataSetConfigurations/sample_organization/2017-05-31_mSEM_scMS109_bk_100um_v01-aniso call returns:

{
    "quality": 0,
    "layers": {
        "color": {
            "brightness": 0,
            "contrast": 1,
            "color": [255, 255, 255],
            "alpha": 100
        }
    },
    "highlightHoveredCellId": true,
    "fourBit": false,
    "interpolation": true,
    "renderMissingDataBlack": true
}

In my opinion the answer should have an alpha value of 80 and the color you set in the settings, shouldn't it?

@youri-k Could you please have a look at this issue/bug? I think it has something to do with the backend.

[youri-k]

@MichaelBuessemeyer I found the bug and fixed it 🙂
Additionally, I found something interesting: When you open a new dataset, you get the current dataset default configuration, and instantly respond to the server with this default config as your new user config. When the default config changes afterwards, you won't notice the change because your user config takes precedence. I don't know if this intended, but it's the same on the master. My feeling would be to only save a new user config, if the user actually changes his dataset settings and not just copying the default config at the point of first opening the dataset.
/cc @philippotto @fm3

[MichaelBuessemeyer]

@youri-k Awesome :D
Thanks for fixing the bug.
According to your thoughts, I think this is not an issue to discuss in this PR as this is how the behaviour has being for a long time. Could you please open a new issue for this?

[philippotto]

Cool, then we can merge this tomorrow, right? @youri-k Can you take care of this (plus migration)?

[normanrz]

Nice! Now we need to put this into voxelytics (and maybe wkcuber)