On Wed, 2007-11-28 at 21:47 -0600, Les Mikesell wrote: > Dave Ihnat wrote: > > On Wed, Nov 28, 2007 at 08:23:56PM -0600, Les Mikesell wrote: > >> I can't recall a case where split with a byte count followed by a cat of > >> the generated files didn't reproduce the input. What can possibly go wrong? > > > > Maybe I'm showing my age, but Unix split used to be only for line-oriented > > text files. Has the current version for Linux become binary-compatible? Using such tools on Linux and other OSes which are not distinguishing between text files and binary files has never been a problem. The problems occur on OSes which distinguish text files from binary files, such as Windows based implementations (Cygwin, MinGW). On these OSes using *nix's text-file utils can have unexpected side-effects. A classical example from the past had been "sed" on Cygwin. Cygwin had treated sed as "designed for text file usage" (opened sed output as "text"), which had caused running sed to perform \n->\r\n expansion when being used on a "non-binary" mounted filesystems and not to perform \n->\r\n expansion when being used on "binary" mounted filesystems. Ralf
