switch to tidync from ncdf4? about rerddap

[sckott]

@rmendels pinging you here to make sure you're aware of the tidync details.

@mdsumner sorry to maybe re-hash the conversation we had in slack ... I want to check what you think about the approach I've taken. Here's the code that uses tidync https://github.com/ropensci/rerddap/blob/tidync/R/zzz.r#L67-L74 - where the input var x is a path to the netcdf file. I know in slack you mentioned maybe doing something like purrr::map(tn$grid$grid, ~activate(tn, .x) %>% hyper_tibble()) where tn is the output of tidync(x). I don't understand enough what the right thing to do here is, but between you two i'm hoping we can pick the best approach here for the most people

[rmendels]

@sckott @mdsumner - For my packages what I need is where the downloaded netcdf file is stored (and that is still there), the packages are indifferent to how rerddap then reads in and melts the data. For users of the rerddap package, if the end result is different in significant ways than in the present version, that could break scripts. I know next to nothing about the internals of tidync, and the real question is using tidync to ingest and 'melt' the netcdf file, rather than using the code you have in the present version of rerddap to do the same, and the best way to use tidync to do so. For the latter question I have no idea, for the former as I said, as long as the end result is basically the same, it doesn't affect anything, sort of one of those internal black-box details that the user doesn't know about.

[sckott]

thanks @rmendels

[mdsumner]

I'm similar, didn't know the context when answering "how to suck out every grid as a dataframe?".

tidync just lets you choose, array output, df output, or just updated meta-print (what slices would be made if read triggered .. )

[sckott]

Thanks both for the input.