On Wed, Mar 31, 2010 at 12:06:03PM +0100, David Cantrell wrote: > On Wed, Mar 31, 2010 at 08:46:19AM +0200, Steffen Mueller wrote: > > > In another mail, you argued that the maintenance burden of code that > > nobody touched in the past ten years can't be all that high. But maybe > > it should be? If the purported replacement modules have been changed in > > the meantime, isn't there a non-zero chance that the old tools have bugs > > going unnoticed? Nobody is going to even look at them to check. I think > > that is a pretty strong reason to NOT ship them in new releases. > > Talking of old buggy tools, how about File::Copy? It's in core, not > deprecated, last updated in 1996, and known to be buggy. For those who > rely on its current behaviour, it would be trivial to CPANise. IMO it's > a far better candidate for deprecation than Class::ISA is. > Last updated in 1996? I beg to differ. Borrowing some code from a cow orker, I get: $ git blame -wM Copy.pm | perl -anle '/\b((19|20)[0-9]{2})-/ && print $1' | sort | uniq -c 254 1996 4 1997 18 1998 6 1999 8 2000 80 2001 15 2002 6 2003 2 2004 28 2005 9 2006 52 2007 23 2008 143 2009 3 2010 > See, for example, this post from 2005: > http://abigail1.hates-software.com/2005/09/21/0692681a.html While I maintain that 'system "cp", "file1", "file2"' is to be preferred over "File::Copy::copy", some of the issues addressed in the article you quote have actually been fixed. AbigailThread Previous | Thread Next