Data Sparrow
Known boundaries
Data Sparrow is deliberately a table workflow, not a spreadsheet clone, hosted database, or guarantee that every file fits in memory.
File size and memory
Local limits depend on measured browser and device behavior. A large source, expanded workbook, join, pivot, or comparison can exceed available memory. The app will offer sampling or an explicit API option where supported, never a silent upload.
Spreadsheet fidelity
XLS, XLSX, XLSB, and ODS imports expose values or cached formula results only. Formulas, macros, formatting, charts, external links, and embedded objects are not executed or preserved. XLS, XLSB, and ODS are import-only.
Inference is not truth
Column types, delimiters, encodings, locales, headers, and date formats can be ambiguous. Inference is labeled and raw strings remain authoritative until you explicitly convert them.
Pseudonymization
Masking, hashing, and token replacement reduce exposure but do not automatically anonymize a dataset. Re-identification risk depends on context, auxiliary data, secrets, and the chosen transformation.