JSON, Avro, ORC, Excel and some others are read whole, and datui asks first when that would pass read.memory_warning (1 GiB). Parquet, CSV and Arrow IPC are read lazily. How each format is read covers each one
datui asks before downloading
A web file, or a bucket object in a format that cannot be read in place, is copied to the temp directory first: How each format is read
A sort, query or analysis is slow on a large file
It reads every row the operation needs, not just the screen: Large datasets
A CSV’s first row is data but is read as the header, or the other way around
--no-header, or H on the Info panel’s Schema tab: Delimited text
A file has no extension, or the wrong one
--format NAME; datui --help lists the names: Formats
A file datui cannot read opens as bytes
No reader or format spec recognizes it: Hex view, and Format specs to describe it
Errors, warnings and backtraces go to datui.log in the cache directory:
The log. File an issue at
https://github.com/derekwisong/datui/issues with the log, the command and,
for public data, the dataset.