I’d like to announce TableView.jl, a WebIO.jl based package for inspecting arbitrarily large Tables.jl-compatible tables (DataFrames, Queryverse-tables, etc…).
Multi-column filtering and sorting works for small datasets (< 10k rows) and is completly handled by the frontend. For large datasets (which are lazy-loaded into the UI, so there’s nothing stopping you from viewing tables with arbitrarily many rows) this does not work yet, but could be implemented in the backend.
It might be possible to extend this package’s functionality to include editing cells, but there are no immediate plans for that from my side.
IJulia and Juno* natively display the output of showtable, but if you’re working in the REPL you’ll need to use e.g. Blink.jl to display the table (see screenshot above).
*) Only works with WebIO 0.7.x right now, but I’m working on a fix for newer versions.
Nice! Works very solidly, thanks for all the work you (and Shashi) put into this. One minor comment: I think it’s perfectly fine to go for lazy loading (and I’m impressed by the excellent performance on a large dataset I’ve tested with) but it may be better to disable the filtering and sorting buttons when the data is too big and they don’t work (if there is a simple option for this in the javascript library), otherwise they are a bit misleading.
Great! This was of the few remaining reasons I use Stata for data analysis instead of Julia.
The next major reason is Stata’s great support for panel data. It does lags and differences as a simple L.varname or D.varname once you’ve specified the ID and time keys.
Anything that can render WebIO contexts, which means HTML+JS+WebSockets, basically.
I don’t think he wanted a more lightweight option, just that the filter and sort buttons are disabled unless they actually work (which I definitely agree with). You can look into BrowseTables.jl for a pure HTML solution though.
