At $work we use a variety of Perls. Strawberry Perl on a Windows machine which does some data munching for accounts - it is probably an early strawberry perl - not sure. Elsewhere Perl is installed all over the place on a variety of hardware including Linux 32/64, Solaris Sparc/Intel, AIX, HP-UX, OSx and loads more lesser known/used. On Linux machines mostly the system supplied Perl is used and on other operating systems it is either perlbrew or built by ourselves and installed into /usr/local/perl. These non-windows installs are mostly for testing or to do some specific job e.g., an internal web site.

For the current project I'm working on we use 5.16.0 installed via perlbrew but in all honesty we could upgrade pretty much whenever we want. Mostly, recently, we've only upgraded when hitting a Perl core issue which was not back ported (the last one was a taint issue IIFC) as it is quite a lot of effort to get all non-core modules installed again and tested. Our 5.16.0 contains one patch from an issue I reported a while back ("double free or corruption" "Invalid write of size 4" in File::Glob? and your post reminded me to check what happened to it.

We use a few features from more modern Perl but not a lot. We briefly used smart match then when it changed (between 5.10 and 5.10.1 IIRC) and we got caught out we stopped using it. We use state, // and some regexp enhancements quite a bit (named captures etc). Unicode support is very important to us to anything that in that area gets our attention quickly.

At home I use loads of different versions under perlbrew mostly and also strawberry perl so I can test modules I maintain. I also have a raspberry pi running a recentish Perl but I forget which version and a nettop box running Linux and Perl 5.16.0.