Testing RHN stuff
Testing testing testing. Fixed an errata fetching bug found in the server code. The new client seems not to be using it, but, since it is "better" than the old client code, it needed to be fixed. NOTE: I hate SQL.
Did lots more testing though one never does enough.
Fixed the way one of our daemons gets serviced. I.e., added pretty [ok]/[failure] messages upon things like server xxx restart. Added restart! And status options. Neat stuff. Never did this before... kinda a hack, but oh well. It works.
Working on shelve2.py
shelve2.py as mentioned before allows you to have nearly any object as a shelve key. Unfortunately, cPickle doesn't always store the dumps result the same way in every case with the same object. This is bad. I spent some time trying to find the answer... well... I didn't. I looked at the C code (4.5K lines) and decided to send a message to comp.lang.python. I wrote a nice pickle test that illustrates the problem. You can download it here: pickleTest.py.
pickleTest.py illustrates how dumps(<object>) and dumps(loads(dumps(<object>))) can yield differing results (with cPickle), though they shouldn't!
National Guard this past weekend
For some time now, I have been in search of a decent tank unit (I moved from Ohio in February). Not only do I need a tank unit here in NC, I need one that has an open Tank Commander spot. So far I have met with much resistance. I am the outsider that threatens established NCOs who are gearing up for promotion. *sigh* There is no way around it... I am being threatened with administrative demotion, though I think I can fight it. We'll see. Luckily my old unit is giving me plenty of time to work things out. Oh well, at least I got to play with tanks this past weekend.