Open cboettig opened 4 years ago
Thanks for filing this from that thread! I think generating a blank CSV and/or data.frame
, paired with a really solid help page would be a good start. A few rows with examples actually sounds really solid and we could basically remove them when set_
ing them.
The Shiny app is nice, I hadn't seen it in a while. I see some areas for improvement there. Might be a good second pass at this?
I would like to see a function which supports importing said csv/xlsx template into the attribute fields, without making the user have to go in and specify the codes for each factor.
It's really unclear from the EML interface/docs how users are supposed to create the table for the attributes list. We should probably at least provide a template XLSX and/or CSV file with the headers filled in (maybe some examples too?)
The documents show one such example, but they show it as tribble, which naturally users may feel means they too are supposed to enter it as a tribble instead of entering it in Excel or similar. Also they don't show all possible columns/attribute types anyway. Even so, said table might remain pretty confusing, particularly since most fields apply only to one type of attribute (e.g. it's probably not obvious that a column called
formatString
refers only to dataTime formats...)Ultimately I think the
dataspice
->EML
approach would give a simpler/more concise way of generating basic metadata for an attributeList and other fields.Or perhaps best, there is that shiny app that Mitchell added,
shiny_attributes()
to do just this, but I believe it needs a bit more help on debugging (at least regarding unit / customUnit) before it's ready for primetime?