Put your inplace-edit backup files into a subdir

[merlyn]

I just discovered that the inplace-edit value can now include an asterisk to mean "old file". This means you can now do cool tricks like put your "old" versions of the file into a subdirectory with the original name, not a modified name! Here's a one-liner that puts all the old files into a directory named for the unix epoch timestamp (a 10-digit integer increasing by one per second).

perl -pe 'BEGIN { mkdir $^T; $^I = "$^T/*" } s/foo/bar/g' file1 file2 
+file3 ...

[merlyn]

I've had more than one person write to me in the day since I posted this snippet to say "Randal, you forgot the -i! It's not an in-place edit without that!".

No, I didn't. Setting $^I is the programmatic equivalent to setting -i on the command-line.

Cool, eh?

-- Randal L. Schwartz, Perl hacker
Be sure to read my standard disclaimer if this is a reply.

[Aristotle]

I was personally puzzled at what $^T is for a moment, though I knew about $^I. So the simple form would be

mkdir bak/ ; perl -pi'bak/*' -e 's/foo/bar/g' file1 file2 file3 ...

Or

perl -pi'bak/*' -e 'BEGIN { mkdir($^I =~ m!(.*/)!) } s/foo/bar/g' file
+1 file2 file3 ...

Or maybe even

perl -MFile::Path -pi'bak/*' -e 'BEGIN { mkpath [ $^I =~ m!(.*/)! ] } 
+s/foo/bar/g' file1 file2 file3 ...

*grin*

Makeshifts last the longest.