@skoulouzis, does it make sense to modify the model used in the containerizer service to match the one in the catalogue?
If we keep the current model in the containerizer service, we will need to translate the cell representation to match the catalogue's model in the frontend (after receiving the response from NaaVRE-containerizer-service/containerize and posting the cell to NaaVRE-catalogue-service/workflow-cells/). This is doable, but we might create confusion by having two formulations of the cell model across different services.
Notable changes:
types is removed and its values merged with inputs, outputs, params or secrets. Eg.
added: description,virtual_lab(probably should be added in the frontend),container_imageandsource_url` (the source folder on GitHub)
removed: node_id, task_name, all_inputs, base_image, notebook_dict and original_source are removed
chart_obj is not in the catalogue model, but it would be nice to still return it for now. We don't have TS code to generate it in the frontend, but we have the Py code to generate it in the backend.
(owner, created, modified are automatically added upon creation in the catalogue)
In the catalogue, we tweaked a couple of things about the cell model (see implementation and sample serialization).
@skoulouzis, does it make sense to modify the model used in the containerizer service to match the one in the catalogue?
If we keep the current model in the containerizer service, we will need to translate the cell representation to match the catalogue's model in the frontend (after receiving the response from
NaaVRE-containerizer-service/containerize
and posting the cell toNaaVRE-catalogue-service/workflow-cells/
). This is doable, but we might create confusion by having two formulations of the cell model across different services.Notable changes:
types
is removed and its values merged withinputs
,outputs
,params
orsecrets
. Eg.becomes
param_values
is removed and its values merged withparams
. Eg.becomes
base_image
→base_container_image
description,
virtual_lab(probably should be added in the frontend),
container_imageand
source_url` (the source folder on GitHub)node_id
,task_name
,all_inputs
,base_image
,notebook_dict
andoriginal_source
are removedchart_obj
is not in the catalogue model, but it would be nice to still return it for now. We don't have TS code to generate it in the frontend, but we have the Py code to generate it in the backend.owner
,created
,modified
are automatically added upon creation in the catalogue)