@fredericcaplette makes sense, it's like preferring in line JSX over returning JSX w/ certain components via conditional. TIL!
@jgdoyon1 how will we buffer this response in GKG side? that could be quite expensive and maybe even a DDOS vector?
@dgruzd sounds good!
Michael Angelo Rivera (dc5008d6) at 18 Mar 15:19
Address self-review comments from MR feedback
updated to use papaparse
@jgdoyon1 yeah, I'm in between on this one. I'll leave it out for now.
@munthank changed it to filter entities from what I remember in our last discussion - let me know what you think the intended behavior should be. We can also handle follow-up MR.
Michael Angelo Rivera (0ed71fc4) at 18 Mar 15:11
Address self-review comments from MR feedback
Michael Angelo Rivera (ad2767c5) at 18 Mar 15:08
Address self-review comments from MR feedback
I saw console worker errors in the browser here - so this fixed it
This may require a proto change so this might be needed as a follow up
self-review: these should come from the backend (graph schema API call) - we should not be hard coding these
@munthank I don't think we should call this "Search Entities" unless we are actually performing a real backend search. If the intended behavior is to filter entities on the existing page, we should call it "filter entities".
@munthank I'm curious where these width and height segments come from? https://threejs.org/docs/#SphereGeometry
I don't necessarily like we're computing the composite key in the frontend, but it is what it is for now.
@jgdoyon1 @michaelusa maybe the GKG web server should serve composite keys as node ids?
Note: this data structure comes from gitlab-org/orbit/knowledge-graph!490 (merged) and gitlab-org/orbit/knowledge-graph#243 (closed)
This is untyped because we want to be flexible with the API structure initially.
We will later put GraphQL, however @fredericcaplette, one thing we will need to think through is the fact that this API will be able to return full block file content blocks from code. I'm unsure whether GraphQL can handle that payload size. Shouldn't block this MR but something to investigate, so I will cross-link to gitlab-org/orbit/knowledge-graph#263
self-review: could these be separate utilities?
self-review: do we not already have a csv package?
This could later be abstracted into loadInitialData, depending on the entry point (e.g., loading data when something comes from a group, project page, pipeline page, etc.) in the product, but this is better as a follow-up (cc @munthank).