`outputinfo(filename)` for inspecting output without loading data

[glwagner]

Another utility that I believe is needed is a function that displays the information in an output file. For example something like

julia> outputinfo(filename)

which displays things like

• the names of the fields
• information about the fields? location, indices, size
• the grid
• the number of output snapshots
• the output times

anything else?

Originally posted by @glwagner in https://github.com/CliMA/Oceananigans.jl/issues/3793#issuecomment-2395088313

[glwagner]

@ali-ramadhan you mentioned that the show method for FieldDataset has something similar. What does this look like?

[navidcy]

Suggestion: If the filename loaded is, e.g., a _part1.jld2 then it would be of interest to detail that and whether this is part1 out of X?

[glwagner]

Yes though I don't completely understand --- if the user writes "part 1", would they not know its part 1?

Maybe this should work if "part1" is not provided? It would also be nice, separately, if FieldTimeSeries could construct a continuous time series.

[navidcy]

Hm... true. Perhaps this is a different discussion, but probably we want an output without any _partX ending in the series of output files with splitting? (It's definitely a different discussion and when that's settled we can revisit the outputinfo(filename)! -- What you suggested above sounds good!)

[ali-ramadhan]

@glwagner It's this show method: https://github.com/CliMA/Oceananigans.jl/blob/03b8acf4f378eeefdb5e79ceeafcf29fa711e94c/src/OutputReaders/field_dataset.jl#L61-L72

and looks something like this:

FieldDataset with 9 fields and 0 metadata entries:
├── v: 865×421×1×5761 FieldTimeSeries{OnDisk} located at (Center, Face, Center) of v at /home/alir/test/simulation_surface_slices.jld2
├── S: 865×420×1×5761 FieldTimeSeries{OnDisk} located at (Center, Center, Center) of S at /home/alir/test/simulation_surface_slices.jld2
├── w: 865×420×1×5761 FieldTimeSeries{OnDisk} located at (Center, Center, Face) of w at /home/alir/test/simulation_surface_slices.jld2
├── T: 865×420×1×5761 FieldTimeSeries{OnDisk} located at (Center, Center, Center) of T at /home/alir/test/simulation_surface_slices.jld2
├── Alk: 865×420×1×5761 FieldTimeSeries{OnDisk} located at (Center, Center, Center) of Alk at /home/alir/test/simulation_surface_slices.jld2
├── DIC: 865×420×1×5761 FieldTimeSeries{OnDisk} located at (Center, Center, Center) of DIC at /home/alir/test/simulation_surface_slices.jld2
├── u: 866×420×1×5761 FieldTimeSeries{OnDisk} located at (Face, Center, Center) of u at /home/alir/test/simulation_surface_slices.jld2
├── pCO₂: 865×420×1×5761 FieldTimeSeries{OnDisk} located at (Center, Center, Center) of pCO₂ at /home/alir/test/simulation_surface_slices.jld2
└── CO₂_surface_flux: 865×420×1×5761 FieldTimeSeries{OnDisk} located at (Center, Center, ⋅) of CO₂_surface_flux at /home/alir/test/simulation_surface_slices.jld2

so it's missing the the output times and perhaps more grid information. But in general it could be improved. I just ended up calling summary(fts) for each FieldTimeseries.

[glwagner]

Is FieldDataset designed so that every field should have the same times, grid, etc? (I guess we can ask the same question for a JLD2 file.) It'd be nice to be able to assume this.

[ali-ramadhan]

It was indeed designed just to provide convenient access to all FieldTimeSeries from one JLD2 file.

This assumption could of course be relaxed as FieldDataset doesn't rely on grid or times, but this is how I've been using it. Same with JLD2 files for me. Surface fields get one file, 3D fields another, zonal slices another, etc.