https://datahubproject.io logo
Join Slack
Powered by
Hey Team, I would like to extend the entity model ...
# advice-metadata-modeling
r
Hey Team, I would like to extend the entity model in datahub to integrate with "Datadoc" in "QueryBook" project. I would like to check if I need to make frontend change in order to present the "Datadoc" entities? (I am a frontend noob)
l
This is great! Please send the proposed model extensions as a RFC/google doc that we can review. You can then contribute the model extensions. We are happy to work with you on enabling the front-end piece. cc @little-megabyte-1074
r
Hey @loud-island-88694 thanks for the reply. Follow this page to submit an RFC, right?
l
yes - that works
r
Not sure how much the model extension will benefit the community, but I will submit an RFC to let community discuss
l
would be useful to know if this can model any notebook integration
r
Maybe some of the model could be re-used by the notebook integration.
another question is I found there are several data schemas have the same info but with different schema name, like 
EditableChartProperties
EditableDashboardProperties
EditableDatasetProperties
, do we want to have something common model like 
EditableProperties
so that future model could just re-use it directly
One more question: a datadoc could have cells. There are 3 types of cell in querybook, text cell, query cell and chart cell. It seems we could model datadoc like dashboard and charts. But the difference is that a cell could only be included in one datadoc unlike chart. I would like some suggestion about if we should model cell as an aspect of datadoc or as an separate entity and define relationship between datadoc and cell
my preference is to model cell as aspect of datadoc, like has 
content
aspect for datadoc and the content is list of cells
l
would wait for @mammoth-bear-12532 to chime in on this ^
r
ack
@mammoth-bear-12532 could you help to have a look at this?
m
@red-napkin-59945: extracting the common pieces into 
EditableProperties
and 
include
ing them from these other aspects makes sense.
thankyou1 1
If a cell doesn't have any identity / life outside of the datadoc, then it seems reasonable to model it as an attached aspect
how are the cells positioned in the datadoc
do they have a coordinate?
r
a cell only lives inside a datadoc. a datadoc is composed of a list of cells. There are 3 types of cells now, text cell, query cell and chart cell. Here is a gif copying from https://www.querybook.org/
our initial goal is to make the text and queries in a datadoc searchable
@mammoth-bear-12532 @loud-island-88694 I am trying to write the RFC for the DataDoc model extension
Do I need to fill every section in the RFC template?
l
no - just focus on the essentials
we can keep it lightweight at this stage and then get into more details on the PR
r
kk
@loud-island-88694 @mammoth-bear-12532 Here is the PR about the RFC. It basically introduced what is DataDoc, why we want to model DataDoc and how we initially model DataDoc
I kind of request a fast first round reviewing to see if community wants this extension or not since it will decide how we execute our plan
4 Views
Open in Slack
PreviousNext
Powered by