https://dagster.io/ logo
Powered by
#announcements
  • s

    schrockn

    11/01/2018, 8:39 PM
    @Ben Gotow did you get a chance to check that out
  • b

    Ben Gotow

    11/01/2018, 9:04 PM
    hey hey not yet - been working on the DAG rendering
  • f

    freiksenet

    11/05/2018, 9:20 AM
    oh damn, this looks so pretty
  • f

    freiksenet

    11/05/2018, 9:20 AM
    @Ben Gotow such a great work 👍
    🙌 1
  • f

    freiksenet

    11/05/2018, 1:42 PM
    @Ben Gotow one note about GraphQL fragments - we are actually sharing one namespace for fragments, so I think we should name all of them 
    ${ComponentName}Fragment
  • f

    freiksenet

    11/05/2018, 1:42 PM
    eg currently you have PipelineFragment in JumpBar
  • f

    freiksenet

    11/05/2018, 1:43 PM
    and it works just because actual fragment that it uses and shares name with loads separately and before that one
  • f

    freiksenet

    11/05/2018, 1:43 PM
    also another note I have - if you are requiring component fragment within other component's fragment, only do it if you are actually using the component
  • f

    freiksenet

    11/05/2018, 1:44 PM
    otherwise it's better to repeat the data requirements
    👍 1
  • f

    freiksenet

    11/05/2018, 3:59 PM
    @schrockn @Ben Gotow https://reindex.daily.co/hello
  • b

    Ben Gotow

    11/05/2018, 4:00 PM
    oh hold on lemme get chrome
  • s

    schrockn

    11/05/2018, 4:33 PM
  • b

    Ben Gotow

    11/06/2018, 2:20 AM
    hey @freiksenet if you get a chance, I think that there may be some misaligned graphql types on develop now after your cleanup PR - I think there are some stray files in src/types that make it work, but some of those aren’t being autogenerated any more. (importantly PipelinesFragment)
  • b

    Ben Gotow

    11/06/2018, 2:21 AM
    Also just to make sure I’m following the right pattern… if 
    <PipelineExplorer />
    defines a 
    PipelineExplorerFragment
    , and that is rolled up into it’s parent’s fragment, should it’s input prop be an instance of it’s own type (
    PipelineExplorerFragment_Pipeline
    ), or the encompassing type declared in it’s parent PipelinePage? (
    PipelinePageFragment_Pipeline
    ). I think it’s the former right?
  • f

    freiksenet

    11/06/2018, 7:58 AM
    it should be instance of it's own type and ideally it should always be the fragment type
  • f

    freiksenet

    11/06/2018, 7:58 AM
    not sub-type
  • f

    freiksenet

    11/06/2018, 7:59 AM
    so PipelineExplorer should accept PipelineExplorerFragment from the parent
  • f

    freiksenet

    11/06/2018, 7:59 AM
    I think we hsould remove all props that are subtypes and if there are multiple inputs to the component that are fragment derived, then have multiple fragments
  • b

    Ben Gotow

    11/06/2018, 8:22 AM
    sounds good - I think we currently pass pipeline solids as a subtype down into the graph / sidebar
  • b

    Ben Gotow

    11/06/2018, 8:23 AM
    out of curiosity, does this approach result in people making a single React component that defines a fragment and renders the whole thing, with sub-render functions, etc., since creating a nice tree of components means creating a whole tree of repeated graphQL fragments?
  • f

    freiksenet

    11/06/2018, 8:24 AM
    but they won't be repeated, because parent usually just needs to use the child fragment spread inside it's own fragment
  • b

    Ben Gotow

    11/06/2018, 8:24 AM
    I guess this might be a somewhat unique case in that we pass almost the entire solid object all the way down to an SVG component several layers in
  • b

    Ben Gotow

    11/06/2018, 8:24 AM
    ahh gotcha I see
  • f

    freiksenet

    11/06/2018, 8:25 AM
    basically what I want to ensure is that type system can check that a) we pass all the props we request with the fragment b) we minimize the chance of "accidentally" achieving this, by only using top-level fragments
  • f

    freiksenet

    11/06/2018, 8:25 AM
    like b) can happen if we have a parent that requests all child fields by accident
  • f

    freiksenet

    11/06/2018, 8:26 AM
    I feel only using top level doesn't eliminate it, but makes it less likely
  • b

    Ben Gotow

    11/06/2018, 8:26 AM
    yeah - b) is pretty much what I was doing I think
  • b

    Ben Gotow

    11/06/2018, 8:26 AM
    just taking advantage of the parent component’s fragment type
  • f

    freiksenet

    11/06/2018, 8:26 AM
    I think Relay actually checks that stuff for you, maybe it's possible to do some magic for that in future
  • f

    freiksenet

    11/06/2018, 8:26 AM
    but I think it's good enough atm
1...545556...185Latest