The amount of code required to accomplish this task depends greately on what kind of file he's dealing with. Using one file instead of two is not always simpler/easier to code.

If files that you're dealing with are small; i.e., config files, etc., it's probably easier to slurp them into memory, perform modifications, truncate the file, and then write data back -- no temp files needed.

With larger files and variable-length records, it's easier to work with temp files. If you attempt to update a variable-length record file without having a temp file, the code will get very ugly in no time. Sometimes, however, you just have to bite the bullet and do it; i.e., when you deal with huge files and you can't spare space for a temp file of the same size.

If you're working with large fixed-length record files, it is almost always better and more elegant, in my opinion, to work with the original file without creating temp files.

--perlplexer