Friday, March 12, 2010

NYC! PAR::Packer!

For a city that never sleeps, morning sure rolls around early in Manhattan
--me
I'm spending a week back in NYC. James is getting married on Sunday! It's been too long since my last visit, I almost forgot the feeling of awesomatude.

Working out of the RubiconProject:NYC office is reinvigorating me. It's so small and full of energy. There are only like 11 desks (soon to be 13). Think particle in a box. Eating lunch around the big round table is very familial. 16th and 7th ave is a nice location with big windows and high ceilings. Last night I walked to Union SQ, the east village, and west village and meat packing district. Mike B keeps the office running smoothly and puts on a solid music mix. Len's got 5 client facing meetings today, it's great to see that kind of flow in the pipeline. zoom!

Waking up at 9am local (6am brain time) is taking an interesting toll on my ability to concentrate on code. Recognizing this I'm working on different problems like using pp to produce a self contained perl script to simplify pushing to the hadoop cluster. I eventually pushed through the biggest hurdle which was ... not verifying my initial conditions.

Turns out someone had some strange code in a standard location that conflicted with the local modules on the box I was using for the build. Private namespace collision 3 names deep? Really? That and DBD::mysql didn't get included into the PAR and/or .pl even with the -B flag to pp. I just forced a yum update install perl-DBD-mysql install on all the target boxes to bring in mysql, DBI and DBD:::mysql ( which is supposed to be in our base builds), and now it's working.

pp is in the PAR::Packer distribution. Give it a spin if you ever need to bundle up an app with all its modules as one bundle. Outputs any of PAR, self contained .pl, or self contained binary that doesn't require perl to be locally installed.

No comments: