> From: Gavin Smith <[email protected]> > Date: Sat, 3 Oct 2026 22:38:01 +0100 > > The claim that files should CRLF line endings should always be treated the > same as LF line endings appears to be based on wishful thinking rather than > reflecting common practice.
On input, it happens automatically on Windows when using the default text-mode I/O (as opposed to binary I/O, which one must explicitly request when opening a file for reading): the libc input functions remove the CR characters from CRLF pairs, leaving just newline visible to the application. (The reverse happens by default on output: CR is emitted before each newline.) I'm mentioning this in case some people in this discussion think texi2any must do something with input files to accept CRLF EOLs on Windows. It doesn't have to, unless it for some reason uses binary I/O.
