These are chat archives for frictionlessdata/chat

Dec 2016
Yuri Astrakhan
Dec 11 2016 00:30
@rgrp, hi, so i got a bit confused about nullability. My goal is to simply mark any columns as "nullable", similar to SQL
the "default" values has a somewhat different semantic meaning - its like adding an automatic SQL collation on top of a null value - something that might be useful in general, but not required in the first pass
also, BIG ANNOUNCEMENT: tabular data has launched on all of Wikipedia. I am still in the process of making a demo, so we haven't announced it widely yet. is some data for an example
Rufus Pollock
Dec 11 2016 09:40

@nyurik i’m not sure i understand. You want to have columns that are null right? That is exactly what missingValue definition is for. As the spec states

By “missing” we simply mean null or “not present for whatever reason”. Many datasets arrive with missing data values, either because a value was not collected or it never existed.

The missing value stuff is not about collation. If you are reading frictionlessdata/specs#97 we are addressing the second of two options that the Stiivi identifies:

Other use is to tell the analytical/datamining software which values are considered as empty (NULLs). For example a string field with a content string "NULL" or "(empty)" will be converted to real NULL value.

@nyurik very exciting to see this go live on Wikipedia!
@nyurik you might also be interested in the work we are doing on a visualization / view spec attached to data packages that builds on vega / vega-lite - something for us to discuss in the near future
Rufus Pollock
Dec 11 2016 10:01
@pwalsh have you seen the updated proposal re metadata translations here: frictionlessdata/specs#42 - - interested in your thoughts. I think we are close to something that is “ready to merge"
Paul Walsh
Dec 11 2016 13:51
@rgrp yes, i’m happy with it. Just some comments on languages property.