Line endings and encodings.
WHAT THE LINE ENDING PROBLEM IS
Windows uses two characters to end a line; Unix uses one.
WHAT THAT CAUSES
Scripts failing with obscure errors Configuration files not parsing Stray characters at line ends
WHAT THE CLASSIC SYMPTOM IS
A script reporting a bad interpreter, naming something that clearly exists.
WHY
The invisible carriage return became part of the interpreter's name.
HOW TO SEE THEM
The file command reports them, or view the file showing non-printing characters.
HOW TO CONVERT
A dedicated conversion tool, or a substitution with sed.
WHAT CAUSES IT
Editing a file on Windows and uploading it Copying through a program that rewrites line endings
WHAT PREVENTS IT
Editors configured to preserve Unix endings Transferring in binary mode
WHAT THE ENCODING PROBLEM IS
Text stored in one character set and read as another.
WHAT IT PRODUCES
Accented characters and symbols appearing as nonsense.
WHY IT MATTERS FOR DATABASES
Importing a dump with the wrong character set corrupts text permanently.
WHAT TO SPECIFY
The character set explicitly, on export and import.
WHAT TO USE THROUGHOUT
A single modern encoding.
WHAT TO CHECK AFTER ANY MIGRATION
Records containing accented or non-Latin characters.
