After buying an Apple iPhone 3G recently, I quickly discovered the reports about short battery life were all true – the Achilles Heel of an otherwise cool device. Making it through a day on just one charge was a challenge. Plugging in overnight was not optional. Once, I’ve even run out of power *despite* having started on a full charge – in frustration, I bought spare charger cables for the office, the backpack, adapters for the car. [Sigh]

Then I found Solio.
Basically, its a battery that you can charge using built-in solar panels or from a USB port on a laptop. Once charged, you can carry it in your bag, and then use it to charge an assortment of small electronics as needed.
 
Pro:

• Seems to charge quickly, and in various light levels.
• Holds that charge for at least up to a couple of weeks of being carried around in a bag, and still works as hoped.
• Brain dead simple to use.
• When fully charged, it can recharge a dead iPhone, and a GPS receiver, and a bluetooth headset, with still some extra juice left over for at least some part of another charge (still experimenting with power limits).

Con:

• The device has two connections on it; one for charging the battery (power in), and one for using the battery to charge other devices (power out). Personally, I’d prefer to have the power out connection (and maybe even the power in connection?) be standard mini-USB. All the electronics I use have cables that connect directly to USB/mini-USB, so this would have made sense to me. Instead, the Solio uses some proprietary form of pseudo-mini-headphone jack connection. They give you a cable to convert from that to USB, which you can then plug your USB-to-device cable, so it does work fine… but….well… it just seems to me that it could have been neater if there were fewer cables to carry around. Not sure how much money they make on selling the extra cables, and not sure of the design/manufacturing consequences of using a slightly larger mini-USB, but this feels to me like a v1.0 nit.

Summary: I’ve used this for about a month, and really like it so far. There’s something refreshing about a device that works as advertised, right out of the box! The nit about connection sockets is just that, a personal nit. Of course the real test will be later this year, when I go camping in the desert. Lets see how that goes! 🙂

Last week we shipped our 5th funnelcake release.

Ken has been leading us doing one funnelcake release per month for 3 months now. Ken can best discuss the significance of all the stats/metrics which these funnelcake releases make possible.

However, I’m more interested in the process. I stumbled across this diagram in bugzilla. To me, this diagram accurately captures how different groups within Mozilla are coordinated do a specific funnelcake release. Having the bug dependencies match the reality is really important. In a time-crunch, with lots of people switching hands, it does make it easy to quickly see what needs to be done next. Each time we do a release, the process gets just a little more refined, and more efficient.

(click through to live bugzilla graph – with options at bottom to change how graph is drawn.)

Hope the diagram helps explain some of the mechanics?

John.

(Personally, I think its a little weird how arrows go “towards” the dependent bugs, not “away from” the dependent bugs. This feels like the “wrong” way to me, but hey, thats just my perspective/style. Certainly the dependencies and interconnected-ness of the bugs is accurate. We havent done a formal post-mortem yet, but based on informal discussions last week, each funnelcake releases feels like an improvement on prior releases; we should be able to improve on this release also.)

To help give a rough outline of how 2008 looked like in RelEng, I thought these numbers were interesting :

At 2008-01-01

• 4 people (bhearsum, joduinn, nthomas, rhelmer)
• 3 code lines (FF2.0.x, FF3.0.x, and a few straggler machines on FF1.5.x)
• 89 machines

At 2008-12-31

• 9 people (alice, aki, bhearsum, catlee, coop, joduinn, nthomas, as well as part time: armen, lsblakk)
• 5 code lines (FF2.0.x, FF3.0.x, mozilla-191, mozilla-central, tracemonkey) all with full sets of build/unittest/talos machines
• 269 machines

I tried to figure out how many releases we did during the year. According to https://wiki.mozilla.org/Releases , we did 40 releases during 2008. However, there’s some other releases I remember happening which never made it to that list, so I actually don’t know what the fully accurate count is. Regardless, “at least 40 releases”, that’s a lot!!