Jeff's Austin Blues » Uncategorized

SxSWi Day 5 – Bruce Sterling

Jeff — Tue, 16 Mar 2010 23:51:49 +0000

Another amazing, dense, rich monologue from Bruce Sterling. A great way to end SxSWi. Hopefully a full video will eventually make its way onto the Internet. Most of the great lines are on Twitter, #brucesterling. Time to rest and start working through the backlog of URLs I e-mailed to myself, Twitter hashtags for sessions, and my notes.

SxSWi Day 4 – Fun Tweet from Results Only Work Environment

Jeff — Mon, 15 Mar 2010 20:12:56 +0000

robert1er: Managers are attached to presence of staff more than performance of staff. We need teddy bears for when employees are gone.

Old Bad Guy, New Suit

Jeff — Sun, 20 Dec 2009 00:49:08 +0000

My laptop has been getting occasional disk errors when doing backups lately. This is not good! I don’t really want to have to buy a new laptop. However, an old memory finally surfaced.

Several years ago, I upgraded the hard drive on a desktop. It was bigger, faster, and required the new 80 wire cables. However, it kept throwing “Media Check Failed” errors. I googled, asked questions, pulled my hair. Finally someone enlightened me, media checks are checksums on blocks of data transferred from the disk controller. Failing media checks are usually caused by bad cables. Replacing the $10 UDMA-100 cable with a $40 one from a major manufacturer solved the problem.

I finally noticed that the current setup is throwing media check failed errors. The backup disk is an external USB hard drive. It is plugged into a USB hub. And it is fast, faster than the built-in hard drive! Just maybe, the USB hub is fine for mice and printers, but maybe, just maybe, marginal with a very fast hard drive.

For the past week, I’ve plugged the hard drive directly into the laptop and there have been no errors. It looks like I’ve found the bad guy, even disguised in new clothes

If It Has a Timer, It Needs a Pause Button

Jeff — Sat, 28 Feb 2009 16:57:03 +0000

My electric toothbrush and the flashcard (jMemorize, very nice) program I am using to learn Spanish both have timers. The toothbrush has a builtin 2 minute timer with beeps every 30 seconds to switch quadrants. I knew 2 minutes suggestion for effective brushing and thought I was fairly close; turns out not to be the case. jMemorize has a selectable lesson size either by time or by number of cards. Two 17 minute sessions works better for me.

Both have a problem handling interruptions, e.g. the phone rings or my wife asks me a questions. If you are creating a program or device with a timer, have a way to pause it. Only hermits have the luxury or reality of no interruptions, for the rest of us, most things need to be interruptable.

System Monitoring = Real-time Regression Testing, AKA Continuous Integration

Jeff — Thu, 05 Feb 2009 23:06:41 +0000

I’m noticing a convergence between the tools and widgets I’m using for system status monitoring and for software testing. There are red/yellow/green backgrounds for specific components, RDD strip charts to track status over time, hierarchies of sub-systems/components, etc. I wonder how much I can leverage the tools and expertise on one side to improve the other.

Manufacturing Applied to Software Development Makes Me Nervous

Jeff — Mon, 26 Jan 2009 20:39:53 +0000

There is a rash of recent software development methods (e.g. Lean, Kanban) that are based on manufacturing, usually Japanese. Manufacturing is doing the known over and over, trying to squeeze the last nickel out of the recurring costs. Software development is largely exploring the unknown. There aren’t repeatable, consistent tasks. Many of these methods strike me as attempts to replace expensive, talented individuals with cheap, generic labor. And sacrificing long-term viability (and costs) for short-term gain.

There are consistent, repeatable parts of software development and where they exist they have been pushed into compilers, scripts, and code generators. Valuable effort has gone and is going into squeezing another couple of percent performance out of the optimizer in compilers. This is so successful that few projects generate assembler/machine code by hand any more. The repeatable parts of well run projects are automated, compiler back-ends, scripts, automated builds, automated tests, etc. Much more cost-effective than hiring poorly paid, poorly educated, unexperienced people.
Or not bothering with documentation, comments, and clean code because it doesn’t provide immediate “customer value”.

Less is More

Jeff — Tue, 02 Sep 2008 16:15:25 +0000

I am preparing for a talk at the Lone Star Ruby Conference on Ruby and Rails on NetBeans, a IDE written in Java. It’s only ten minutes long and so a bit of a squeeze. With practice and dropping all the “You knows…” and such I can shorten it, but I still have a 20 minute talk. Listening to a podcast last night by the Common Craft folk, I was reminded that it is more about what you leave out than what you leave in. It was reinforced this morning by a video on the Guardian’s site on New Orleans and Hurricane Gustave. There is no voice over, just the sound of the wind and pictures of trees, sofas, and people being pushed about by the wind. The last third is President Bush boarding a plane. Again no voice over, just the whine of an idling jet. Powerful.

Linux Promoters Chasing the Wrong Angle

Jeff — Thu, 14 Aug 2008 14:51:53 +0000

Two blog entries muse on what works in the marketplace and why Linux isn’t making huge inroads: Matt Asay on
The Linux desktop, Macs, and barking dogs and Seth Godin on The intangibles. My experience agrees, cool sells, in spades. Linux’s promoters, advocates, etc. are bucking a strong headwind in the anti-business and RTFM attitude. Volunteers rarely have the time, patience, etc. to build an maintain the kind of relationship Seth talks about. The cool Matt talks about may be easier to push, but Seth’s approach has the better payoff.

Programming, Process, & Feedback

Jeff — Mon, 02 Jun 2008 21:46:12 +0000

The original waterfall model of programming process had no feedback loops, i.e. “complete and correct” requirements were handed off to the architects who produced a “complete and correct” system architecture, etc. It quickly became obvious that the various documents were seldom completely “complete and correct” so some rework was required. But the various process “gurus” still make pronouncements like “complete and correct” is still doable and try to pretend that the various feedback loops wouldn’t be necessary if you would just do the process right. It’s pretty obvious that that emperor has no clothes.

Most of the agile processes elevate feedback to a first class part of the process. On-site customer or stakeholder representatives give feedback, unit tests give feedback, if pair programming is used, programmers give each other feedback, etc. Feedback is no longer reluctantly allowed in to accommodate imperfect programmers but is encouraged, built into the process, and pushed to as early as possible. Testing can begin as soon as there is executable code.

Nokia N810 – GPS

Jeff — Wed, 13 Feb 2008 22:38:35 +0000

The GPS receiver in the Nokia N810 isn’t very strong. I have yet to get a position fix indoors or even in a car. Outdoors it does acquire position fixes, but there is a fair amount of uncertainty/jitter. Standing still my “position” wanders about. It is good enough to figure out which block in a city your are on and even which end if it isn’t a tiny block. The maps in the database are low resolution, i.e., line segments, not smooth curves. The Points of Interest (POI) are sometimes out of date (e.g., restaurants that have closed). Only roads are included, no footpaths, so not very useful for hiking with the included maps. (Is there a source somewhere for the USGS topographic maps?)