On Tue, 30 Jan 2007, Jacques B. wrote:
files= for f in * do [ -f "$f" ] && files="$files $f" done ... do stuff with $files ...
If you've got to deal with filenames with whitespace in them you need to be trickier.
I like how this works. But you are right, it does cause a problem for filenames with spaces. As far as cpu time (preceding the command with "time" to get same), we are only talking a few 1/100th of a second difference with the find command. The find command is more robust as it will properly deal with filenames with spaces. Not to mention find will also yield hidden files (i.e. .file) whereas the above won't. All depends on your needs as to which will serve you better.
Note, that find also solves another problem: namely, that if the globbing generates a line length beyond the max for bash (in either the for or the concatenation), then the script will fail. This won't happen with find.
Steve Friedman