| Walt Mankowski on Mon, 4 Nov 2002 18:40:55 -0500 |
|
On Mon, Nov 04, 2002 at 05:45:00PM -0500, gabriel rosenkoetter wrote: > On Mon, Nov 04, 2002 at 03:53:00PM -0500, Walt Mankowski wrote: > > split(1) (at least GNU's version) works just fine with binary files. > > But (and I think we've had this generic argument before), > Sun's/SGI's/whatever's may not. So it's a bad idea to get in the > habit. And dd(1) is more versatile anyway. By that same argument you'd never use any of the nonstandard features in the GNU utilities. > What's more, split(1), by default, divides a file into files of "1000 > lines", and doesn't have a good way to specify a byte boundary, > which is what you really want to specify. (A binary file may simply > not contain a ^J or a ^M anywhere.) GNU's -b flag sort of lets you > do this, but it may make reconsruction later complicated. (Was there > a CR? Was there an LF? Were there both? Neither? Did split(1) > swallow them?) That might be an issue in general, but it this case it will work just fine. copy /b foo+bar baz will reconstruct the file without changing anything in foo or bar. > dd(1) is the Right tool for breaking up files along byte (or disk > block, or whatever arbitrary size larger than a byte you like) > boundaries. I disagree. The Right Tool is whatever lets you get your job done most easily, not whatever works generically on the most different platforms. If you have GNU's split, it's a lot easier to use than dd. Why not use better tools if you have them available? Attachment:
pgpmrgEZPcfay.pgp
|
|