Test cases for a text field

The browser counts the field as filled and the server counts it as empty, or the other way round.

A field that looks empty but is not: required-field validation passes, and the record has no usable name.

Sorting puts the record at the top of every list, and exact-match search never finds it.

Login and e-mail comparisons differ between browser and server; the account is created but cannot be found.

The classic column width: one character short of the limit, exactly at it, and one over it are three different code paths.

One character over the classic limit, which is where silent truncation usually happens.

128 characters but 256 bytes: a field limited in bytes rejects a value the interface said was fine.

The limit of the most common text column type, where crossing it usually produces a raw database error.

Encoding narrower than the full character set replaces these with question marks, and the name is corrupted on save.

The same letter written as a base plus a combining mark compares as different from the single-character form.

A character outside the basic plane in a name field breaks byte-length limits and databases configured for narrow encodings.

A direction-override character makes the text render reversed: the value shown to the user is not the value stored.

Serializers and loosely typed interfaces turn the text into an absent value; the field looks saved and comes back empty.

YAML 1.1 and several config parsers read the unquoted word no as boolean false - and NO is the country code for Norway.

Numeric parsers accept it, and then every comparison against it is false - filters and sorting quietly drop the row.

Often the result of a bug elsewhere, so a real value spelled this way is indistinguishable from a defect.

Quantity, price and count fields that never expected a negative value; often accepted and only noticed at invoicing.

The decimal separator used across most of Europe; parsers expecting a dot read it as a thousands separator or reject it.

One past the most common integer limit - where counters, identifiers and quantities wrap to a negative number.

Above this value browsers and many interfaces stop representing whole numbers exactly, so an identifier returned by the server no longer matches what the user sees.

A date that cannot exist; naive parsers roll it over to the second of March instead of refusing it.

Valid in some years and not in others, so the same input is correct or incorrect depending on the year field next to it.

The same text means March or April depending on convention, and nothing on screen says which was assumed.

Five-digit years break fixed-width parsing and column formats without producing an obvious error.

A value starting with an equals sign is treated as a formula by spreadsheet software; the exported file shows a computed result instead of the data.

A comma inside a value splits one column into two unless the value is quoted correctly at every export step.

Codes lose their leading zeros the moment a spreadsheet treats the column as numeric.

Spreadsheets keep about fifteen significant digits, so long identifiers come back rounded and in exponential notation.

A device name reserved by one operating system; a file with this name cannot be created and the failure often surfaces as a generic error.

Trailing dots are silently removed by some file systems, so the stored name differs from the requested one.

Interfaces hiding known extensions show this as report.txt, so the user sees a document and gets a program.

Two dots mean the parent directory; the shortest possible traversal test.