Recently in Technology Category

In addition to having a thing for Vespas, there's something about the steampunkery of vintage motorcycles that makes them mesmerizing.

Originally posted at advrider.com, I saw this English rally linked from the 2strokebuzz blog.

Some of the pictures on the road are fuzzy, but scroll down to where they're all parked on display in Brighton.

Check out the "steering wheel", and all the squeezy bulb horns
(eg:), the suspension, and "tags".

And it's almost got to be open source maker friendly if this is your repair kit:

I was poking at my server a week and a half ago and wasn't happy with the state of something so I decided to power it off (at the PDU...I mean why have it consuming power and only logically off?) rather than leave it in an in between state while I ate dinner. Turns out it didn't power back on automatically when power is restored, presumably due to bios (mis)configuration. Oops!

Finally had a fellow coop member in California able to press the power button for me today. Strangely it wouldn't go into the BIOS for them and rather than mess with trying to diagnose that (it's been like this for two years and wasn't an issue) I've just left things as they are for now. At some point this machine's hardware is going to die and it'll get sorted out on the new hardware (or virtual machine).

I never did get that bios "feature".

My solar panels haven't quite passed inspection yet (next week?...fingers crossed), but they're producing and the data collection module for the PVPowered inverter is sending data out to the mvpvpowered.com web site.

I want to be able to push the data to my own site though (along with the weather station data). And in theory PVPowered will let me. The PVM1010 module's page states:

Communications options: Standard open protocols that work on TCP/IP such as UDP and MODBUS can be supplied for system users on request.

In the meantime I thought I'd have a look at what the module does. First most obvious step was to point a web browser at the IP my router handed out when I plugged in the module. Nothing. Okay...port scan...nothing! Locked down. So next step was wireshark to watch for packets out of the module. And...after a few minutes I catch the once per fifteen minute squirt of data out to dcgateway.pvpowered.com. Amazingly it's an SSL encrypted session!

While this doesn't mean the embedded device is secure, and in theory it shouldn't much matter if it is just collecting data off the inverter's serial port and it can't actually send signals and harm the inverter, it does show that somebody at PV Powered put some proper engineering thought behind this module.

I'm impressed!

Though now I'm really counting on them to hook me up with a way to get at the data from the module.

I find it really humorous that as soon as the iPhone 3G is unshackled ATT starts sending out regular emails and SMS's with tips for travelling internationally with an iPhone. The email says "Tips to minimize international data charges when travelling outside the U.S." What more tip does one need than to unlock the phone and use a country-local mobile provider?

Since Dan Kegel commented on my blog, I thought I'd confirm the Windows version of SketchUp does work on x86 linux via wine. It's a bit slow at times and goes into deep never-never-land at times only to return and seemingly run at near native speed. But if I'm wanting to do some minor twiddling of a small model while on the go, I do technically have the option on linux with the T61p I'm using.

This is in Fedora 10 with rawhide's wine (1.1.10 though 1.1.11's supposed to be better) and compiz turned off and an Nvidia graphics card. The last point there though comes with a requirement on the Nvidia binary-only proprietary driver, which in effect means you're running Sketchup on an Nvidia kernel with a linux OS shimmed into it...the Nvidia driver is nearly 7MB while the (unnecessarily large for this machine) linux kernel plus all other drivers is combined under 6MB on this machine.

At any rate, wine is certainly an impressive endeavour and it's amazing that it can successfully allow a migration path to linux for even complex windows applications. I certainly used to use it daily until the last non-linux app I had to use came out with a native version.

Now if it could only run Husqvarna / Viking's 4D Software (overpriced, requires hardware dongle and crazily expensive sewing machine) then I might be able to get my parents off Windows too.

I saw this awesome video on CurrentTV this morning. Not that I follow the experimental musical instrument scene, but this is one of the slicker things I've seen lately for any type of embedded device, much less musical.

Laptops are pretty standard on stage at live show these days...things like Rebirth and cd turntables have been around for a solid decade now. But the Reactable...

Just yesterday I was lamenting the lack of good ways to interact with a 3d model in Sketchup. These 3d mice look interesting (and have linux support), though I wonder if they require a helicopter pilot license. There's major promise though if comp sci doctorates are making devices like Reactables!

Oh and the Reactable runs linux. I love the Xubuntu logo on the table as they start to calibrate the projector and camera in the middle of the video.

Awesome.

I've been using Qcad for a year or two in order to do simple 2-d drafting for woodworking projects. It works just fine and I know how to get around in it efficiently. But it's been a bit cumbersome at times.

I'd tried Sketchup briefly two weeks ago when thinking about making the tv stand. But it was new to me and I didn't want to give the time to learn it. Eric and Jason were bending my ear about Sketchup, so this weekend I gave some time to trying to re-draft the minimal 2-d mock-up I'd done for the dining room panelling I've been needing to finish for a looooong time (so I can get on to the cabinet and beams in that room).

This is extremely minimal in qcad and it got the job done, once I'd done some math and figuring on paper. I don't believe in qcad one can group things. That would make it more useful for mocking things up as I try to develop an idea and remove a lot of pen/paper work.

Sketchup on the other hand has groups and components. The third dimension of things is pretty easily pushed and pulled out of 2-d sections, which you can even create by tracing sections from your millwork supplier's photos / web imagery (though some might even have their stuff in the Sketchup's 3d warehouse?). Eric pointed me at a dimensional lumber plugin. With this and the grouping/components it's really easy to virtually assemble a project, get a feel for how it actually looks and try shifting things around to play with proportions.

The main things I've had to figure out were:

• Where/how to click to get the tools to infer what I'm trying to imply (really 3d can be a pain given 2d display and input devices)


• When to use push/pull vs. scale (haven't quite gotten the follow-me tool down yet)


• How to draw guidelines...this was key. In 2d (whether Qcad, pencil/paper or other) for me it's all about drawing guidelines especially when projecting a front to side and top views. The tutorials I'd looked through on Sketchup didn't really talk about guidelines.


• And again where and how to position/click the mouse to get things to jump onto the guidelines/points.


• Figuring out that you can start a rectangle (for example) then select pan/orbit tools to move elsewhere in the composition and then reselect the rectangle tool and it will still be waiting for the second point to finish defining the recetangle.


• To push/pull/edit part of a group/component you can explode it (bad) or edit group/component. That sort of visually drops you into changing just the selected group/component.


• Oh and realizing I could just type distances/dimensions without actually mousing over to the distance/dimension input box and putting the focus there. There isn't a visual hint to indicate the focus is already there and for some reason it looked like the focus was specifically not there to me.

I like doing drafting and am having fun trying to make furniture. Learning this new tool should make me more productive!

Now if only it were available on linux.

It's been on the back of my mind for ages to check if the Peer to Patent project has an RSS feed for new patents added to the system. Today I actually noticed I'd written it on a TODO list at some point and since it would only take a second and my mind was already on it...I checked.

Not too suprisingly there is a feed. Given the ease of adding feeds to a feed reader, the huge value in having a massive, distributed pool of eyeballs helping patent offices and that from what I've heard the biggest weakness of Peer to Patent currently is the lack of reviewers...I thought I'd do my small part by throwing a suggestion out into the blogosphere that people subscribe to this feed.

Last I knew Verizon was still a fan of a tiered-internet. On the other hand they're now saying at least when it comes to blocking copyrighted materials they wont tier things.

Their VP of PR specifically has said, “We generally are reluctant to get into the business of examining content that flows across our networks and taking some action as a result of that content."

I'm not sure how they can say that and at the same time be for a tiered internet. Unless they envision that as discriminating against traffic by source and destination only and not content? But some of the other quotes in the NYT article make it seem like Verizon may be moving more towards accepting that they're in the business of selling pipes and the more and fatter pipes customers want because of a thriving internet means more business for them.

For somebody interested in learning about linux memory management it can be hard finding documentation. There are a couple books like my coworker's book (that's starting to get dated) Understanding the Linux VMM and the stand-by from O'Reilly, Understanding the Linux Kernel. There will always be random other little things scattered around the web like the occasional posts on lwn.net, usually around some particular patch or set of patches working its way through the community. Of course there's Rik van Riel's linux-mm wiki and the linux-mm mailing list. These are mostly focused though on kernel implementation details and there is a bigger picture than that. This fall there's something new to add to this list.

A series of linux memory articles was published on lwn.net by Ulrich Drepper (glibc maintainer, RedHat employee). He recently also made the paper in its entirety available on his web site.

One of the unique and interesting things about this set of documentation is its scope. He starts with low level descriptions of RAM circuit implementations and works his way up through computer architecture details and on to OS and application software. At each level there are complications to work around the limitations of other layers.

Some thoughts after reading it:

• With large numbers of cores in a system, how do you deal with scaling the cache coherency? Snoop/broadcast can't scale forever. Intel's Terascale docu references directory based approaches. The thought of the cache itself having NUMA style latency differentials and programming for those makes my head hurt.


• Regardless of explicit awareness in the application, a fancy system resource scheduler might consider working set size, instruction mix, cache sizes, bus speeds and hit/miss rates in order to migrate both processes across processors and their associated memory around within the system. But are there actually schedulers anywhere that are this sophisticated (mental note to self...google scholar search)? I doubt it.


• Besides the rare people who know enough about their workload to know disabling hyperthreading (in the BIOS?!) is the right thing to do, would normal people actually choose to leave half of their Intel quad-core chip unused in order to give a full L2 to the one in-use core in each of the quad-core's core-pairs if given the option? These are complicated trade-offs and users aren't informed in an end-user-consumable fashion to make a smart decision. Even if they were they wouldn't.


• I'd never heard typical processes don't fill the TLB before a context switch and the resulting TLB flush, but this is logical (although I'd have liked to have seen some stats to back this). It's obviously even worse if the OS requires a flush on mode switch between user and kernel which the paper seems to imply is normal, but for the processors/OS he's talking about though this shouldn't be the case. The first issue though says a lot about why we have such small TLB's (and TLB coverage) today. The best I'd heard was a bigger TLB would be more expensive, but I've never seen good justification at a simple cost level, much less weighing other factors like selective invalidation cost vs. simple/full flush, larger pages, likelihood of full TLB and so on. TLB-fullness is at least easy to measure for a workload. If cost were the only issue I'd expect to see occasional bumps to the TLB size as a chip matures or its manufacturing process shrinks, but there's no point in an expense of the workloads don't fill the TLB. It does make me wonder though why haven't more processors done better TLB tagging and selective invalidation instead of full flushes? Perhaps the processor architects have simply bet on larger page sizes for workloads that can fill the TLB. It'll be interesting to read Mel's thesis.


• And larger page sizes are getting easier. Despite the problems outlined in the paper, events got a bit ahead of Ulrich this past fall (in a good way) and the hugetlbfs mount's pool of huge pages is getting to be (see Mel's blog) better able to handle the issues he points out. Not that it's a panacea. But progress is good.


• So we're seeing that low level microarchitectural interactions matter. But the average developer does not comprehend them. And the tools to analyse things like cache misses and bus contention are not easy to use so app developers don't stand much chance. Maybe perfmon2 (if it makes it mainline) eventually enables fancy tools that help a users and app dev's quickly glimpse the computer architecture impacts of their code, but this all seems a long way off. In the meantime with oprofile you've got to be quite savvy (ie: know what counters your specific hardware has and how many and probably write wrappers to multiplex the multiple events across the few counters in combinations that are valid) to learn things. The feedback loop needs to be shorter and easier to manipulate.


• The NUMA section is particularly interesting because Ulrich and Andi Kleen (maintainer of libnuma and numactl...the standard linux userspace support for application NUMA-awareness) had a pretty massive difference of opinion on how kernel NUMA abilities should be exported to userspace a few years ago. I haven't seen much from Ulrich on the topic since. Andi's libnuma became the code shipped by most distros, but it never was integrated into libc. And in my opinion what we have today is quite complicated to use and Andi's not super responsive or open to changes on that front. Ulrich doesn't get into this, but the end user implementation facts are important here. My understanding is that many people today don't bother to try to tune because it's too hard to get right in static situations much less dynamic ones or portable code. This will be exacerbated by the increase in NUMA hardware. Not too suprisingly though the section 4 NUMA reader comments are pretty sparse. This isn't easy stuff to grasp.


• And section 5 just gets that much thicker. Very useful discussion and pointers on code and data layout. And while the user knobs here are perhaps easier than figuring out optimal memory/cpu/node placement for a multithreaded workload, this stuff is similarly not super approachable to a run of the mill developer.


• Hmmmm wait a second! Speaking of Drepper v. Kleen...Section 5 references a libNUMA (yes, upper case) and section 12. What's this? Different than Andi's libnuma? And more subtle dissing of libnuma (yes, lower case) in section 6.5 and reference to an alternative. Then tucked away in the appendix is something that hints at the difference of opinion between Andi and Ulrich. Namely a section 12 and "libNUMA Introduction". So he has done an alternative NUMA library for linux. Argh. This hardware management stuff is complicated enough, but now we're growing complexity in userspace too (eg: two numa libraries, cpusets, and other things). Ulrich's own employer (RedHat) ships Andi Kleen's libnuma (see numactl package), but not Ulrich Drepper's libNUMA. This just makes it that much harder for the end user or programmer to do something intelligent about the complex memory issues Ulrich is trying to highlight.


• As with most anything like this, the examples are sometimes a bit unclear. Without specifics on the hardware clearly spelled out statements like, "But for four threads the numbers for the last test show that it is almost not worth it to scale beyond two threads..." combined with the earlier statement of, "This means the prefetch traffic together with the write-back traffic is pretty much saturating the bus when four threads are used," aren't exactly helpful beyond showing that things are already fairly ugly (which is useful in itself). But this doesn't help a user select a system or a programmer design their application.

All in all the paper's well written and provides useful information and poses equally useful questions and problems for consideration. The above comments probably highlight that I'm seeing a huge problem in the usability of this stuff...

It's simply way too hard today for somebody to find the info they need to tune their system and for application developers write portable code that is tuned/tunable. There's been much talk of "autonomic" computing for years, but we're not anywhere close to containing the complexity of optimal performance.

In my last entry I didn't really describe what all the kids had to do this year and I should have.

Each year the FIRST LEGO League competition has a theme. Last year the theme was nanotechnology which was a real challenge for fifth graders. But this year's theme was much easier because issues of power consumption, reduction and alternative energy production are all around us day to day. The kids started out somewhat familiar with the concepts already and it was much easier for them to research and comprehend the problem space.

In addition to move visible aspect of building LEGO robots which compete on a table to manipulate challenges in a proscribed way, the teams also get points in the competition based on judging of the technical merits of their robot mechanism and software designs, judging of an interview on their teamwork skills and judging of a presentation.

The presentation this year had three components: go out in the community and perform an energy audit, analyse the audit data to find ways to reduce energy consumption and adopt alternative energy sources, and finally return to the community building to present the findings and proposals. LEGOtricity audited the headquarters of Medical Teams International here in Tigard. Their primary observations were that the heating of the office space was poorly balanced, the windows and concrete walls and were not sufficiently insulating, their water heating seemed inefficient (old boiler far from usage) and there was a substantial amount of electricity being used for lighting and appliances. MTI in particular was interested in their suggestions for tank-less water heaters and removing one of three tubes in the flourescent light installations throughout their cubicles. The team also researched solar power, but found MTI would need something like a $4million installation of panels (which they did calculate could fit on MTI's roof). The bummer is that MTI is a non-profit so they don't pay taxes and can't take advantage of the massive solar incentives available today.

To that end I really want to thank Dan Crowell (Excalibur Solar, LLC) for answering the kids questions about solar electric. After going to the NW Solar Expo earlier this year, I've been thinking seriously of trying to get a photovoltaic installation on my roof and need to give him a call so see about spec'ing something out. I greatly appreciate his willingness to interact with some kids in his local (Tigard/Portland/Oregon) community to help them learn.

This link came across my feed reader. It's not quite clear where it's going as only the first two of five parts have been released as of today. But the Black Hole looks like a really cool store to tool around in for a while. Their eBay store doesn't appear to work, but it would be so much more fun to see the place first hand anyway.

Poking further I've found I didn't have archive mappings post MT upgrade and have created them.

And I re-undid their silly default file permissions.

I see they've changed from using underscores to dashes for multiword files/directories. So there's a bunch of old underscore named content that's been regenerated with dash named content. Seems like yet another pointlessly annoying change. I supposed if I delete my old ones I possibly break old links and need to tell apache to do some url rewriting.

On the bright side I figured out that the missing photo thumbnailing was my part...somehow I lost something I needed and it wasn't MT being brain damaged as I assumed.

And I am starting to find the info I need in the MT documentation as I stumble through problems and am starting to understand how MT is meant to work. I still think they need a better set of initial user documentation that describes the big picture and how different pieces parts of the system are meant to work together.

I still have template and archive and asset tinkering to do, but am making progress.

When looking at system performance it is often useful to get concrete numbers on what the system is actually doing. Oprofile can give you interesting details like listings of every function called, the call graphs and timing information for a live workload on a running system. When the profile shows frequent calls to something like spin_lock() or up_read() though you're going to need to do further analysis...

While much of userspace application development is just starting to take advantage of multithreading, kernels have done so out of necessity for a very long time. Whether user or kernel space, when you have multiple sequences of code that may be simultaneously running and accessing shared data you need to be able to synchronise their activity in order to insure correct operation.

That's where locks (or more formally "synchronisation primitives") come in. Perhaps ideally you desire or naively expect perfectly scalable code for infinitely many-way computing and therefore expect not to see locking primitives in your profiles. But scalability and the day's common workloads are complex, moving targets. So trade-offs are made in locking implementations. For example, originally when the linux kernel began to support multiple CPUs, there was one giant lock to protect anything and everything. Today in the 2.6.24 kernel there are over 17,000 call sites for the spin_lock() function and over 4,300 distinct locks those call sites acquire. And spin locks are only one of many synchronisation primitives available!

In the past, lockmeter was the analysis tool to have in your toolbox when you needed to explore linux kernel lock performance. But it was an out-of-mainline kernel patch and as is invariably the case with such things...lockmeter was frequently not available for the kernel you were trying to debug.

Over the last few years though much debugging code has made its way into the mainline 2.6 kernel, including a framework called lockdep which allows verification of lock dependencies at run time to help catch broken code (synchronisation is notoriously hard to get and keep correct) in the wild. The basic infrastructure is some code and data structures that track which locks are acquired and the context of the acquisition (E.g.: in which functions, with which other locks already held in the call tree above, and whether IRQs are disabled). It's a fairly obvious extension then to also allow for collection of statistics such as how long one had to wait to acquire a lock and how long one held the lock.

This in essence is the lock statistics (aka lockstat) feature which Peter Zijlstra implemented and which was initially merged into mainline as of kernel 2.6.23. A number of useful bugfixes have since been applied in 2.6.24. The obligatory description of the feature is in the kernel source's Documentation/lockstat.txt file. To enable capture of lock statistics, simply build your kernel with "CONFIG_LOCK_STAT=y" on supported architectures.

As of 2.6.24, the x86 (both i386 and x86_64 subarchitectures), arm, mips, s390, sh and sparc64 architectures have the required dependencies for lockstat. In addition to the obvious dependency on lockdep there are also dependencies on the architecture supporting irqflag and stack tracing. There are early support patches[1, 2] that enable all of these for the ppc64 architecture, but these were not ready in time for 2.6.24.

Using lockstat is extremely easy. The file /proc/lock_stat can be read to view the current lock statistics at any time. It is a text file with minimal formatting, including a header and stanzas of data per lock being tracked. Due to the amount of info provided, the output tends to have very long lines (almost two hundred characters). You'll want a terminal a bit wider than the standard 80 characters when using lockstat and CSS that doesn't chop long blockquoted or preformatted lines when displaying output on the web, E.g.:

lock_stat version 0.2
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
                              class name    con-bounces    contentions   waittime-min   waittime-max waittime-total    acq-bounces   acquisitions   holdtime-min   holdtime-max holdtime-total
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

                             dcache_lock:         14739          15215           0.74         266.84       15388.18          47671        1344622           0.46         592.19      562514.34
                             -----------
                             dcache_lock           3470                  [] _atomic_dec_and_lock+0x2a/0x48
                             dcache_lock           1162                  [] d_alloc+0x13e/0x18c
                             dcache_lock             86                  [] shrink_dcache_parent+0x27/0xd8
                             dcache_lock             14                  [] d_rehash+0xe/0x32

...............................................................................................................................................................................................

              &inode->i_data.tree_lock-W:          6810           6990           0.50          32.42        6680.31          12568         129595           0.53          93.77       57411.23
              &inode->i_data.tree_lock-R:          3052           3080           0.71          24.25        2856.64          94357         924633           0.60          76.59      421768.49
              --------------------------
                &inode->i_data.tree_lock              4                  [] add_to_page_cache+0x24/0x76
                &inode->i_data.tree_lock           9602                  [] __do_page_cache_readahead+0xe3/0x210
                &inode->i_data.tree_lock              0                  [] find_get_page+0x14/0x47
                &inode->i_data.tree_lock              0                  [] test_clear_page_writeback+0x31/0x84

...............................................................................................................................................................................................

Simply doing something like 'cat /proc/lock_stat > somefile' allows you to record the statistics at a point in time. Since the file is "just text" though normal system commands come in quite useful. Something like:

watch -d 'grep -e \: -e "class name" /proc/lock_stat | head -11'

would show you the top 10 locks (by contention count), plus a header for easy reading, update the output every two seconds and visually highlight any changes in the updated output.

The above sample output demonstrates the rich detail lockstat gives. For each lock you get a stanza of key statistics, a break out for reader and writer sides of the lock where applicable, and a list of the top four contending call sites in the kernel (memory address and symbolic name). The ten statistics listed per lock class are:

• con-bounces - The number of contended lock acquisitions which involved cross-cpu data.
• contentions - The number of times a lock acquisition attempt had to wait.
• waittime-min - The shortest wait time for a contention.
• waittime-max - The longest wait time for a contention.
• waittime-total - The total aggregated time spent waiting to acquire the lock.
• acq-bounces - The number of lock acquisitions which involved cross-cpu data.
• acquisitions - The number of times the lock was acquired.
• holdtime-min - The minimum time for which the lock was held.
• holdtime-max - The maximum time for which the lock was held.
• holdtime-total - The total aggregated time during which the lock was held.

These statistics also hint at some of the complexities of lock design and implementation (for more information see for example this paper or others which cite it). Not just the simpler factors like the amounts of time for which locks are held or contended matter. The amount of inter-CPU bus traffic, cache pollution or cache synchronisation stalls generated also matter.

And this is one of the cruxes of using lockstat. Lock primitives are often implemented using architecture specific, hand-tuned assembly language. These functions are typically then compiled inline as well. In order to collect statistics though the lock functions must become more complex. So lockstat itself perturbs the system performance by adding a variety of overheads. While SGI's lockmeter could be compiled in but disabled so as to only take a minimal performance hit unless actively profiling, lockstat is always on. Either way the runtime overhead of lock profiling can be quite significant. Lockstat overhead ranges from 20-30% on some popular synthetic benchmarks.

As with most kernel debugging tools, getting debugging information is often only one of the easier steps in the complicated process of engineering a solution. Lockstat's inclusion in mainline means there is now a very easy to use way to get at detailed kernel locking information. Hopefully that in turn gives you a little time back to explore the more interesting intricacies of performance scalability.

My Irish coworker, linux vm hacker and all around smart dude Mel Gorman has started a "semi technical" blog! Most linux development happens via mailing lists, IRC chatting or the occasional face to face, but it's a positive development that people are starting blogs/wikis to track the state and direction of things in a persistent coalesced fashion.

Up until this weekend I've had a large rambling collection of computer hardware "that might be useful" but which wasn't getting any meaningful use by me. I'd been planning to recycle/donate it somewhere somehow and finally got organised, backed up a few things, wiped my hard drives, filled up the Subaru (to the brim) and drove over to Free Geek.

I am seriously impressed by their operation! They're way huge compared to what I expected, had lots of volunteers and a huge stream of people dropping off things. Their thrift store has gobs of reasonably priced recycled systems and accessories. And they're all about linux (Ubuntu specifically) and OpenOffice to keep prices down for their target demographic.

I was a bit sad though to see the genuine Hercules monochrome ISA card and amber monochrome CRT sorted into the recycle pile instead of reuse. But then there probably aren't that many ISA slot sporting motherboards in a chassis big enough to hold the card desired for use by some random nerd like me as a server console. A volunteer said their museum of old computer stuff has had to become picky as they get so much old crap that they'd run out of room.

But it feels good to know some of the stuff (5 computers, 4 monitors, 3 UPS's, and a lot more) should find new homes and if not be recycled responsibly!

It's coming to light that Comcast appears to be interfering with Lotus Notes traffic, in addition to other packet types. This isn't surprising and is perfectly in line with a company that wants a tiered internet so people "don't get things for free". But what if they actually were to put packet transfer rates on the open market and let Microsoft (creator of IBM's primary groupware competitor) and IBM duke it out. Instead getting people to pay twice for service, they could actually extort many multiples of that.

Sure it's a conspiracy nut accusation to seriously think Microsoft would be behind this, but the what-if is the interesting part. It's a much better business model. Why as the consumer and producer to pay a rate on the service they want, when you could inflate that price by allowing the competition to pay for anti-service?

I love to bash EMI because they were at the forefront of corporate attempts at stifling the interactions and subsequent creative possibilities in a networked world when they forced OLGA off the net in 1996. There's no coincidence between Radiohead's contract with EMI expiring, Radiohead shocking the music industry by offering it's upcoming album direct to listeners at whatever price they're willing to pay (or 40 pounds if you want it on physical media including all sorts of fancy extras), and EMI's new private equity owner saying they must embrace digital music or die. But with that guy being rumoured to be planning to sell EMI to Warner and EMI's long history with fighting modernisation, I'm not expecting much here beyond words and further attempts at mildly warping their business into something internet related...no fundamental shift in how they do business.

A couple weeks ago I went to the NW Solar Expo in Portland. I was really struck by how solar is a commercial reality today. The expo was all about how to get solar electric or water heating today. It still takes governmental assistance to make it financially reasonable, but that's a smart investment in the future in a number of ways.

This weekend we were in Bend and they were having a tour of local homes that demonstrate solar and other green building options. Unfortunately we weren't able to do the tour because we didn't have time and it looks like we missed the similar local tours here last month.

This winter I'm going to try to spec out a solar electric system for my home usage, get quotes to figure out the up front price and estimate the cost in the end after all the incentives have rolled in. And presumably I'll then need to figure out whether I can get one of the state's energy investment loans to cover the up front investment. After a few years and all the incentives it doesn't sound like it has to amount to an unreasonable cost, but there is a substantial up front cost before the incentives.

IEEE's Spectrum magazine has an article on this topic. It highlights California's push, but at the end mentions that Oregon has just started some major funding as well. For our state it really makes sense as investment in fostering local high-tech industry and energy independence, not to mention the environmental positives.

This weekend the geek press is all over new terms of service AT&T has rolled out which basically say they will cancel your connectivity if you use it to do anything on the net which they deem disparages their company. This is a brilliant example of why we need net-neutrality. It's like a free-speech protection against companies who would censor us.

I wouldn't be surprised if similar were in other carriers' terms of service...Did I mention how much I really love my Verizon FIOS internet service? Seriously it is actually good. It'd be really even more super if Verizon were a fan of network neutrality instead of a tiered internet.

AT&T's basically given a sweet demonstration of what the lower tiers of a tiered internet will look like. Presumably if you pay enough (eg: corporate customer) and you are disparaging them, they'd prefer to keep your business and thus try to right the problems. But the little guy just looses.

It still sticks out in my head that I had a professor in the mid-90s who was talking about programming and threading and asserted that practically nobody in the world knew how to write multithreaded code. At the time I didn't figure it mattered so much. But with in a couple years I was writing linux kernel code and dealing with concurrency and reentrance issues and realised that it's not exactly trivial to write good parallel code. It's slowly become clear over the years that most software engineers just haven't bumped up against this in their day to day work.

With the push to multicore chips and increasing parallelism in hardware even in commodity hardware, the ability for code to exploit that parallelism becomes very important if computers are to effectively get faster year to year. Given that average coders don't know how to do this...

At OSCON last month I was interested to learn about Intel's Threading Building Blockslibrary to help people write threaded code.

And today I noticed there is similarly an IBM Software Development Kit (SDK) for Multicore Acceleration, which is Cell specific.

I understand chip manufacturers' business inclination to emphasise optimisations on their own hardware. But this sure seems like an area where a standardised set of hardware agnostics abstractions would be very beneficial.

I upgraded my laptop from Fedora Core 6 to Fedora 7 quite a while back and I'm still paying for it. The most painful part was that F7's install CD apparently doesn't support upgrading (although maybe the DVD does). I suppose I'm too used to distros doing the right thing and finding the install on the partition I point the installer at and not doing the right thing with respect to updating. I definitely told it not to format the disk. Regardless it didn't do an update and did do a format. Which really sucked for the random little things I've not thought to backup at all.

Like the power management scripts I'd customised. I've been beating my head against F7 on this for a couple months now in random spurts of frustration at the state to which the machine wakes. My wireless wont work without serious beating and Compiz has left lovely artifacts all over the screen.

Finally this evening I realised my scripts weren't executable. For at least the config variable setting one I thought it was just sourced and didn't need to be. We I have working is via the two files below (plus a chmod +x on them).

It seriously annoys me though that the distro wouldn't have some of these basic things done by default/automatically to get common hardware devices working with their suspend/resume script foo.

Arguably from the Department of You Can Fund A Study To Prove Whatever You Want, but there is now a study claiming that fair use not just adds economic value, but actually adds more to the economy than copyright!

Maybe someday there will be acceptance of the reasoned arguments that holding IP too tightly hurts us, especially in the digital age.

My first job out of college was working in IBM's disk business in San Jose so I'm always intrigued to hear about the latest storage innovations. Last I'd really heard they were big on a MEMS based technology called Millipede, but I haven't heard too much about that in a while. Yesterday a NYTimes tech article made the rounds regarding research into another micro-mechanical type of storage IBM researchers are playing with.

The article variously claims it's a long way from being commercialised and could replace disks and memories in three to five years. Who knows if or when any such scenario might actually occur.

The most interesting thing in this article though isn't the time to market, but the seek time promised by this technology. Disks have rotational delay which means we have data seek times in the range of milliseconds, which is an extremely long time compared to the rest of computer technology. Even solid state memories have pretty long access times compared to the speeds of CPUs. Masking these delays adds all sorts of complexity to the design of the broader system and algorithms.

This research prototype though boasts the possibility of single nanosecond seeks. That would be massively better than disk could ever achieve and is even an order of magnitude better than DRAM today.

Whether it's this or another invention, it's sure fun watching how Amdahl's Law drives the innovations that allow us to at least match the types of scaling indicated in Moore's Law if not outright clobber it.

The iPodTouch's format made me immediately wonder how much of its core was shared with the iPhone. engadget noticed that some of Apple's press showed the iPod with a BlueTooth icon and there is a BlueTooth chip confirmed within it, but disabled. Assuming a bit of hackery, the WiFi enablement and a BlueTooth headset I wonder how long until somebody makes VoIP phone out of their new iPod. I wonder if Apple's exclusive deal with AT&T precludes Apple doing an iPod VoIP deal?

If the press release were dated a day earlier I'd have been sure it was a joke. But it appears the anti-market EMI corporation may be the first major to start getting with the digital future. It might still be cheaper to just buy the physical media though...depends on whether they do $9.99 albums DRM free too.

And it is a bit odd that they'd have the DRM versions at all. Is $0.30/track enough to discourage somebody who wants to pirate music? No. Are people wanting to save a few cents a track seen as implicitly pirate-prone? Must be. The majors sure relate to their customers in strange ways.

Details on Intel's 80core CPUs continue to materialise. The way the press originally read for these last summer they sounded more researchy and less producty, but as this article points out things sound increasingly like their terascale architecture may be more than just crazy research.

Unfortunately it doesn't include new info on the memory side of things.

One of the first things that really jumped out at me was the clock arrival time diagrams. The chip's cores operate mesosychronously. I found this nice analogy on mesosynchronous versus (a)synchronous:

In the field of mesodynamics, the term mesodynamic refers
to the middle ground between classical physics and quantum
mechanics. By mesosynchronous real-time systems we mean
those that are in the middle ground between

• Totally synchronous – in the sense of having only
  static, periodic, time-driven (i.e., TDMA-like)
  activities (or at least such activities are the only
  ones considered important)
  
• Totally asynchronous – in the sense of having only
  dynamic, aperiodic (not necessarily even sporadic),
  event-driven activities.
  The derivation of “mesosynchronous” from
  “mesodynamic” reflects that: synchronous real-time
  computing, like classical physics, is comparatively well
  understood; while asynchronous real-time computing, like
  quantum mechanics, is still comparatively poorly
  understood, and seems to require a paradigm shift on the part
  of both the research and the practitioner communities.

They do note that clock skew across the die gives some power benefit (consumption less spikey). The (many) other power related details show how performance/power relationships have become a huge issue lately.

I was surprised to stumble on this today. The actual write-up is a good read...thoughtful and written in a way I'd expect a large portion of Apple's critics (eg: esp. the techno-illiterate politicians) as well as their customers to be able to fully understand. And Jobs squarely plants the problem in the court of the big labels (for those who didn't already clearly get the bloc power they hypocritically wield)!

This is exactly the type of message that you'd expect to deflect the European political pressure on Apple around its DRM. And exactly the thing to increase pressure on the majors to catch up with reality. Or the thing to increase political pressure in a more appropriate place (eg: how about these ideas for a start) towards helping the majors get a clue.

Maybe I was on the leading edge in witnessing EMI's attack on OLGA (over a decade ago already?!), but I think the rest of Apple's billion downloaders are starting to get the situation and are beginning to see the anti-market, anti-competitive, anti-creative state of copyright this DRM facade masks.

I can't wait to watch how this plays out! It made the tail end of Marketplace today and I'd expect it prominently in the press tomorrow.

The one thing I was a bit worried about on the iPhone was how open a platform it would be and how much OSX it would be...and it looks like there's reason to worry. I suppose if it's got enough OSX that I can open a console window and run ssh I'm mostly happy. But one of the killer things about Palm was (is?) the openness, developer community and the resulting huge amount of useful third party apps. I find it hard to believe that the device's OS couldn't sufficiently isolate things to the point that a misbehaving app wouldn't be able to bring down the cellular network or otherwise cause major problems. I think this is more about Cingular and the entrenched cellular providers clinging to their business model and wanting to ensure a steady stream of controlled, billable events on their network.

An iPhone on Sprint's WIMAX network with a VOIP client would be a cellular industry game changer, assuming the rumours of Sprint making that service essentially like a wireless ISP (ie: here's your IP and bandwidth, use it as you like) come true.

The cloud is coming and devices will be connected.

Or maybe 2 RBI home run even.

The iPhone covers my desire for a combined media player, phone, low end camera and PDA. And good battery life. I'll be buying one of these. Especially since the wifi should mean I'm not getting reamed by ATT/Cingular data service. Can't wait to see the reviews when they start getting into people's hands and how they truly work.

And the AppleTV (not sure how to write that in text without the Apple icon)...No need for them to buy out or partner with Tivo now that ITMS has so much traction. I'll likely go for this device although 720p isn't ideal and I'd also need a machine on which to run iTunes. But given I'd pay about a grand a year to DirectTv for a stream from them with HD and Tivo support (plus nearly that up front for an HD Tivo) and I only watch a few hours of TV a month...I should be buying Apple boxes instead and especially be buying into Apple's ala carte model. I always wanted ala carte channel pricing for TV, but I watch so little, I may as well pay per show. It's got to be a lot easier and likely cheaper than MythTV given how little I watch TV.

I haven't done much of an analysis lately, but for quite some time most of the spam hitting my mailserver originated in Taiwan. I'm curious if the fibre outages from yesterday's earthquake off Taiwan will make a noticeable dent in spam for a few days?

A while back a couple co-workers suggested tomboy for note taking. I was actually looking for a little pop-up reminder application (eg: "bing bing...it's two o'clock, stop typing and go to that meeting") and there are tomboy plugins for things like that. But I didn't want to run a notes management widget and just wanted a reminder widget.

So around rolls the end of the year...it's one of the times when I try to do a little organising and tidying up of loose ends and thoughts. And towards that end, with a lot of random notes all over the place needing some better organisation, I remembered the tomboy recommendation.

Tomboy is a perfect match for me. It's a very simple application that's trivial to learn. It's like a mix between stick-notes and wiki for your desktop. I love it! And I bet others would too if they knew it existed and gave it a spin.

I'm not sure I've felt much of a longing to be in the Bay Area since moving to Portland, but I have to admit that I really miss hearing about things like this and the ability to just drop in and get involved.

The concept of an open source mobile phone is really attractive. I'm pretty unhappy with most of what's on the market in terms of both handsets and service agreements. I would love the opportunity to hack together the type of handset and features that _I_ want...Even if that handset starts out as or even always is mostly a bulky add on to my laptop.

I'm still looking for something to replace my Palm Vx. I'd like something similar size, that's also a phone, has a VGA-ish resolution camera, has good battery life (ie: a number of days when not actively in use) and which has a general purpose operating system and the ability to run arbitrary application of my own authorship or selection. Basically my Palm Vx with a camera and phone. But not so large as the Treo's. And not so crippled by crappy telecom company business plans. I want my phone service to be a pipe over which I can do anything I want...like an ISP. Which brings me back to "just" carrying a laptop and finding WiFi. Not as elegant of a solution and not as technically fun.

After upgrading to FC6 a month or so ago there've been a few quirky things happening on my desktop's display. Today I was finally forced to resolve them.  And in the bargain I happened to get AIGLX and compiz working.

AIGLX is essentially some GL extensions to X that allow for fancy graphic things to happen. Thus far it's mostly been games and maybe scientific visualation work that's used the real power of high end graphics cards. Compiz is the default window manager for use with the GL enabled X servers. Compiz has forked and the resulting "Beryl" looks pretty slick as well. Together these all bring linux's UI up into the eye candy realm of OSX Jaguar (and presumably MS Vista).

This stuff's all fairly new and probably disorients many users more than helps them. Still it is interesting to see people actually pushing the desktop UI paradigm some.

I was inclined to think Oracle's support of RedHat linux was just Larry being bent out of shape about loosing Jboss. Comments from an aquaintance at Oracle didn't do much to disprove that. They're not likely to make any money off of this and providing better support than RedHat should be difficult. Putting out a distribution isn't easy in the first place. Beyond that, replacing the distro's packages with your "updated" packages is liable to wreak all sorts of problems, especially in the enterprise space where there are for instance closed source storage drivers built against very specific distro packages. Or where all the vendors certify their own support on various specific configurations that aren't likely to include Oracle's linux version(s). So neither the business or technical ends of this one have much up side.

And Novell in bed with Microsoft? On the surface partnering on selling licenses could be simply the logical conclusion to Microsoft approaching the linux community this past summer on better virtualisation interoperability.

What makes all of this wierd though is the indemnification, patent protection, cross licensing stuff. I am not a lawyer, but I've seen enough to understand that IP law and business is very complicated. Throw curve balls into the mix like the GNU General Public License and a huge number of copyright holders and it's only going to be more complex. Not to mention the possibilities around GPL v.3.

There must be hoardes of lawyer people trying to sort all this out. Given the complexity it's hard to see things resolving cleanly though. Complex problems, giant corporations and lawyers aren't liable to lead pretty, clean solutions. Even if they're smart lawyers like Eben Moglen.

Ugh.

I've still been holding off buying an HD television, not being completely happy with the LCD and plasma technology on the market (although the prices have gotten much better during my wait). A year and a half ago I blogged about Motorola announcing some new tech, but haven't heard anything more of it. Today another new tech is making headlines and this one claims to be coming to market by December 2007. We'll see if that actually happens, but it reads like they have manufacturer commitment to the supposedly superior technology. I'm a bit sceptical since the news is lacking in real detail and is apparently tied to a stock listing (what happened to quiet periods?).

Slashdot's got two good links on the DRM fromt today.

First, Yahoo's actually taking a stand against DRM "protected" music! This is a great quote:

"As you know, we've been publicly trying to convince record labels that they should be selling MP3s for a while now. Our position is simple: DRM doesn't add any value for the artist, label (who are selling DRM-free music every day -- the Compact Disc), or consumer, the only people it adds value to are the technology companies who are interested in locking consumers to a particular technology platform. We've also been saying that DRM has a cost. It's very expensive for companies like Yahoo! to implement. We'd much rather have our engineers building better personalization, recommendations, playlisting applications, community apps, etc, instead of complex provisioning systems which at the end of the day allow you to burn a CD and take the DRM back off, anyway!"

The second article on /. gives a deeper dive into just that. An analysis of DRM and how it has been circumvented.

Argh. Four hour layover on my way to OLS (yeah!) and the Vancouver, BC airport doesn't have free wifi. After paying $15 for a day's use and nearly that much again for roaming cell call to tech support, who stumbled even just to read an url to me and ask what OS I was using (linux) and can I try rebooting (sure right).

Turns out it was a combination of cookies, popups, their website thinking I was connected but the connection was dropping regularly due to signal strength lacking and a UI that wont show you a connect/disconnect option if you're not already viewing it via their popup window. Walking around I can't find much signal anywhere. Lame.

Makes me really appreciate the Portland airport with free wifi and an initial redirect page that shows you graphically which access point you're coming through so you can get closer if you need to, which has never been the case because signal strength was so high.

So far best signal strength is right next to the smokers lounge and it's still weak.

Last week Cringely had an interesting column on the issue of network neutrality.  Personally I believe network neutrality is critical to innovation in much the same ways open APIs and interfaces and the openness of open source and the commons are.

The telcos seem to be bent on returning us to a simple central controlled broadcast-information society instead of that which the internet has evolved into.  The reality is that decentralisation and participation (peer to peer in its broadest sense) is what the internet is all about today.  And it's reasonable to see that changing substantially without net neutrality.

I really like Cringely's use of the phrase "billable event."  It really summarises what the telcos want and need in order to drive growth and increases in profit.  They aren't content to be the conduits that they are.

The interesting twist I see in this is that while the opponents of opensource and the commons like to dismiss these as socialist or communist pipe dreams, a concept like community funded infrastruture allows the infrastructure to sustain a proper marketplace instead of the infrastructure sustaining a centralised, command economy which is what the telcos have had and want to consolidate.

Wired has an interview with some people from the party. Cool to see them continuing to generate press.

This extremely high gas mileage is interesting in as much as everybody in the US is complaining about gas prices but it seems completely bogus. They don't appear to be doing anything meaningful beyond reducing speed, weight, friction, air drag. So sure they produce a high number, but it's not like we're learning anything new or there's real innovation here that will make any real difference. Most consumers still tend to prefer larger, fast, strong looking cars and have bad driving habits, all of which run counter to fuel efficiency. That's what needs changed and it's going to take more pain in the wallet to do that.

After getting beat up for quite some time allofmp3.com have issued a rebuttal to the US political/economic pressure. It's really a shame that this blatant abuse of the US political system by Hollywood is spreading around the world.

When I first was told about allofmp3.com I was a little dubious given the state of the rule of law in Russia. But they appear to be quite on the up and up. I think I might need to start contributing financially to their cause.

I'd be really interested to know how much they in turn contribute to royalty collection agencies. I wouldn't be at all surprised if they contribute more than other "major" players. But presumably the royalty collection people or Hollywood companies are holding that money in escrow pending litigation, and therefore the major labels are in fact the ones not paying the artists.

I forget where I saw this essay/speech linked, but I finally got around to finishing reading it during lunch today. Definitely a thought provoking read.

Mr. Graham makes some interesting points about where innovations are happening. I'm sure big companies would argue they're generating plenty of innovation, but his points about company internal secrecy and the principles of openness vs. secrecy mesh with my experiences at large and small companies regarding IP as well as my personal views on openness being goodness. It combines to make me think he's on to something. Time to buy more small cap stocks?

That just leaves the patent trolls. He seems confident that'll be solved quickly...Here's hoping. In the meantime I guess we're mostly stuck waiting on legislatures?

Saw this in today's news. Considering that Oracle was looking at JBoss and trying to expand their stack and that RedHat grabbed JBoss out from under them it it isn't a big surprise. RedHat's seemed for some time like they get that the money is in differentiation at the higher end as things commoditise and are slowly working their way up the stack. I could see Ellison refusing to consider RedHat though after loosing JBoss to them, but SUSE/Novell/Oracle just sounds complicated. Since SUSE got acquired by Novell their direction hasn't really been clear, where RedHat's been a bit moreso.

Anyway it'll be interesting to see if and what consolidation happens as linux continues to grow.s

Slashdot's running the latest rev on a many years running story/rumour of IBM moving to linux instead of MS Windows on its employee workstations. Also in the comments are a link to heise claiming IBM's discounted Andreas Pleschek's comments. The latter is in German and to me doesn't actually read like a substantial denial. It definitely doesn't say anything about IBM renewing or cancelling any licensing of MS products.

If Herr Pleschek's statement is true you'd have to imagine he's getting a bit of a hand slapping at work today. And if it is true, that's a pretty huge statement in support of Linux being ready for the desktop. I've personally felt for a while that it is roughly on par with Windows and OSX for most typical end user types of workloads, so it's really only a matter of time before its adoption rate goes up especially in corporate environments where it's probably easier to accomplish a switch compared to within the consumer market.

It will be interesting to see if this indicates IBM's Workplace product will be flexible enough to run on any linux distro and hardware architecture or if it'd be poorly ported to the point of only working on very specific configurations. You'd think if it's eclipse and java it'd run anyway, but any real experience with java or typical first ports of applications to linux would indicate otherwise.

Last week an interesting note came through a thread on the linux-kernel mailing list. Basically Ingo Molnar's calculated, given the size of the code in debian, that linux and the surrounding open source software have passed half a billion lines of code. "That makes Linux and OSS the largest man-made science project in history. It also means that the cost of redeveloping the Debian codebase using commerical methods could exceed 100 billion US dollars." Pretty impressive stuff.

And of course there's the obligatory jab embedded at Microsoft.

I saw this on BoingBoing following up on their reference the other day to this story. Is Houston Police Chief Harold Hurtt planning to run for president or something? He seems to be working on outpacing President Bush at trouncing civil liberties.

This is one choice quote: "I know a lot of people are concerned about big brother, but my response to that is if you aren't doing anything wrong why worry about it."

The best part is there's now a bounty up for any video of Chief Hurtt breaking a law!

This should be fun to watch unfold...

As if this wasn't bad enough looking a month ago...now it's just sounding like total science fiction is coming out of Congress. And from my own Senator now!

Granted Boing Boing's got a sci-fi writer posting, but the EFF's info isn't exactly heart warming by comparison either.

Time to start making phone calls.

Google's at it again...There's been so much press after CES that this one must've just slipped through. But the Economist reports that Larry Page announced "...Google Fastfood, a button in car dashboards that delivers instantaneous hamburgers."

Google's clearly become corrupted! No more "make the world a better place"...they're trying to super-size us. But then, people are starting to be less opposed to being super-sized, so maybe it's not bad after all.

Seriously though, an interesting Economist perspective on being Larry Page.

Of course the devil is in the details, but there is some very promising news out today on the issue of intellectual property reform. The current system is being severely abused and is very much in need of change.

Various sources are reporting on the topic. The key points are:

• an open patent review process
• Open Source Software as prior art
• a unified, numeric patent quality index

Apparently the Blu-ray consortium is having trouble with their Advanced Access Content System. Wonder how much this has to do with Sony and their recent DRM problems? Its amazing how so much time and money can be put into something that will ultimately not work.

In the flurry of legislative action yesterday was a hidden gem from the House for the content industry in the form of HR-4569. Not too surprisingly with a huge amount of last minute legislation yesterday Thomas is way behind today so it's hard to get much detail beyond the above mirrored text of the legislation and the public outcry and the MPAA chair calling it "very important piece of legislation."

I had to write my Representative the Honorable David Wu a note in response.

Today there's a Businessweek article noting the rise of LAMP. Linux and the other components of LAMP are beyond up and coming in my mind; they're pretty well established. But there is always something about them and solutions based on standards and open software that makes people say they're not ready yet for prime time or "the enterprise." This article includes a quote about how LAMP's fine for simple things, but for complex stuff you still need Java. I just don't get that.

The P scripting languages are all quite robust. They have all sorts of libraries and modules to enable just about anything you can desire. They're fully capable programming languages. If you have a way to define your data format and standardise communicating data (ie: xml) and a full featured programming language like Perl/Python/PHP...what is there that you can't do?

Going back seven years ago I was interning at IBM and was working on an application that clearly would've done well on a LAMP platform. I could see it would've been simpler to deploy and support and over time extend. But back then it was way too fringe for people to accept it, especially when the corporate direction was to run everything on Domino. I still wish we'd gone LAMP. Maybe today we'd have been able to...

Thinking about including Google's javascript in my webpages last month (1 and 2) has had me thinking more on how complex interactions happen with software and security implications. The Sony mess with DRM is a good example of complex interactions, but this has largely been interactions on a single machine, possibly with a remote attacker. XSS, email, IM and the like bring in a bit of a network aspect. But it seems like we're getting into a new realm where it's harder to quantify risk or actively manage it as code moves onto other peoples' servers.

That Google was able to "fix" by changing code on Google servers a major security flaw in Microsoft's Internet Explorer that their Google Desktop software exposed is fascinating. Fixing desktop code in a third party's software by putting changes on your server that your user doesn't actually have to download. That's complexity and means problems.

It seems like google is all over the news these days, both for what they're doing and also for how they're doing it. This article on how they're trying to get the most out of their knowledge workers is an interesting read, especially in the context of my own experiences of what works well and what doesn't. Many of the things they talk about remind me of things that were done when in worked in the Materials Lab (then IBM now Hitachi) and which I've missed in jobs since then. I actually just got the invite for the annual MatLab holiday putluck (it's been over 5 years since I worked there and I still keep in touch with the people).

I'd be interested to know more about their data drives decisions though. What data and how do they gather it? More often than not I see silly data gathering in the form of status checking using up to as much as 50% of people's time. Or code versioning / bug tracking systems with too many fields that need correctly set so people can track things. I wonder how they balance generating the data on work with generating the work itself?

As soon as the Mac mini came out I had a PVR in mind and was hoping they'd partner with Tivo.

Currently the rumour mill says a media center is in the works and that a deal with Tivo fell apart.

The HD DirectTV Tivo's been priced outrageously and isn't very featureful compared to real Tivo's. If and when I go to HD I'll be looking at new options, so I'm hoping Apple's in the market as speculated.

It's been interesting to watch how this story has progressed into the mainstream press and the reactions, from BoingBoing's commentary in the expected pretty pissed off blow by blow fashion to the newswire articles in the print paper that finds its way to my doorstep each day.

Today Bruce Schneier's weighed in with a nice analysis.

Sure Sony's backpedaling right now, but this whole issue starts to set a scary precedent for how the industry will operate in the future. As Schneier points out these big companies will just come to the plate better prepared next time and with the proper collusion they'll succeed at screwing us consumers. Their lawyers are probably already working on legislation to work through their representatives to make this easier.

I've been subscribed to AWAD for many years and noticed today that they have their own bit of javascript for inclusion in your web site.

This and the google one yesterday got me thinking think I'd never even looked at plugins for MovableType. There's a huge list.

One of these days I'll have to play with adding some stuff to my blog. I still owe Jenn a Gallery though and should update mine to the new major release. And I should update our MovableType installs for that matter. Guess I'm slacking on the sysadmin front.

Sure I added a blog entry with a snippet of code recently and was blindly thinking it did something smart and dynamic and then realised that it was just a boring bit of html code. I obviously wasn't really thinking.

But Google Analytics got me thinking a little more. They ask you to include a little script of theirs on your web site. In return you get some magic info about your site.

All well and good except what does that bit of javascript do? I haven't looked at it very closely, but close enough to tell it's not the most clearly written code in the world. What if it has a bug? Or what if one is later introduced?

This brings up an interesting twist of a possibility...a commercialisation of the cross-site scripting class of security holes. How long until somebody comes up with some sort of commercial spyware XSS? Maybe there are already things like this lurking in our web applications.

Honda's running a world first trial with a consumer renting a fuel cell car from them. Very cool. But with the press coverage and word that the car is a million dollar car, how long until this LA driver gets carjacked?

Given the extent to which open source and free software is taking off in the developing world, you'd think Microsoft would be making a big push to provide extremely low cost versions of their software in these areas in order to have some mindshare at all in a new generation of technologists. But they think, "You can give people free software or computers, but they won't have the expertise to use it"...This from one of their Africa managers. Sure we see a lot of countries with political and other problems in Africa and the developing world, but are the people there inherently incapable of grasping new tools in order to build local technology economies? I think not.

It's always nice to see that the competition fundamentally does not get it, though.

Jenn and I are trying to decide if we should combine wireless accounts and get new phones. But it's a complicated issue: Go with her Sprint or my Cingular or something else? What phones since there aren't really any that match our desires?

On calling plans what we'd need is equally affordable everywhere. The extras I'm looking for look like they'd run as follows per month:

               Sprint  Cingular   Tmobile

Unlimited web:   $10      $20        $6

100 SMS:         $5       $6         $5

MMS:            incl      $3        incl

Jenn's flying back from Raleigh today via Atlanta. Her Delta 1237 flight from ATL to PDX had a scheduled departure of 09:45am and actual of 09:46am. Not bad...one minute late. But on the arrival side scheduled is 11:53am and estimated actual is 12:39pm. So the plane has lost nearly an hour of time in the air? I really doubt that. I suppose it's possible they're flying slower to save gas and money. But more likely is that they're pushing back from then gate and not departing in order to keep good "on time departure" statistics.

What a crock.

I bet they sell a few more lunches than normal on that flight as well. Why not just be honest with people so they can plan accordingly?

Slashdot's run a post about the possibility of Google Office brought up in this blog. The blogger says the business model isn't exactly clear, but I think it is in line with the other things Google's been doing. Given IBM's componentisation of OpenOffice into Eclipse plugins, it seems like the office suite has already been decomposed into building blockes. It's only a matter of time now before they're built back together into interesting new places instead of just a simple standalone office suite for your desktop machine.

Scott sent me a hilarious piece of cardboard from a cigarette package. He and Charlotte are in Sri Lanka with a tsunami relief organisation. I wonder if this is real or a bootleg pack of fags? If they were fake like Mike shoes you'd think they'd have a Marble lable instead of Marlboro. But maybe they weren't trying to make a joke.

There's also another funny picture I snapped at the Alaska Airlines terminal in San Jose back in August and forgot on my camera...

I never knew Google had this feature. Saw it referenced off BoingBoing.

Given things like an ISBN or a UPC they'll return current prices. Driving directions, movie show times, yellow pages...This is everything I wanted from a web enabled phone. But ATT/Cingular's web interface is horrible compared to what I've seen of Sprint's. My girlfriend's got a good deal (ie: unlimited usage) but my net access isn't very cheap. I can get unlimited SMS quite cheap and with this I don't have to navigate a bunch of sites, just send a simple query.

I think I'll really use this.

Marcus Ranum (smart guy and security expert) has an interesting new essay on common falacies in security. This is a good, thought provoking article.

The NYT and others are reporting on a Harvard plan that's being presented to the World Bank. They're pretty short on details. Somehow the cynic in my can't imagine the World Bank is truly interested in shaking up the established business world in order to help out developing nations. Not that it wouldn't help the developed nations as well. It's just that it's hard challenging successful incumbents. Anyway, I wish them luck.

It's going to be an interesting few years as inniatives like this and the new GPL update push the conversation about openness forward.

Last night I decided to move my home router and wireless upstairs to hopefully get a little better signal propogation for the wifi. I plugged stuff back in and powered up the dsl modem and wireless access point and thought I heard the click whir whir of my linux router machine starting up too. But I hadn't powered that on. Instead I was hearing the pop fizzle fizzle of my WAP11 subjected to the DSL modem's transformer which is 12v instead of 5v.

Maybe this is an excuse to upgrade to a G-capable router and retire the old Pentium 1 133Mhz linux router.

It's been interesting to watch the progression of linux over the years as manifested at the LinuxWorld Expo and this year was no exception. The expo outgrew the San Jose convention center and moved up to San Francisco in 2001 as the dot-com bubble burst. That year the show was very sparse and a few slim years followed, but this year continues the trend that was definitely apparent last year...bigger and ever more corporate.

On the very superficial side of things: I actually felt under dressed this year, but maybe that was just the day/time I was there? Also there seemed to be less women on the expo floor than in previous years (the number had been increasing), but more in the booths selling things.

Venture capital seems to really be flowing again. There are tonnes of little companies doing things with linux and open source. Unlike 2000 they showed up with sanely sized booths instead of massive expensive showy constructions.

In the last year or two there was an increasing emphasis on data center stuff like SAN storage and RAID. Last year there was a lot of clustering. Both of those were present this year, but I noticed a lot of people were pitching things further up the stack like software/system install, configuration and management software.

A few years back Ximian had a big booth, but since then I hadn't seen much sign of linux on the desktop being pushed. There was an exception this year in that Steve Mills (IBM Software Group VP) gave a keynote talking about the possibility of linux on the desktop and Lotus WorkPlace being an enabler there. I've been very skeptical of LWP, but I saw some interesting things. A component based set of OpenOffice plugins for Eclipse is cool. The aspects of Domino Designer I used to like appear to still be present. They showed an interesting demo of collaboration status/event passing beyond simple IM presence (eg: shared files in a document library and who's actually working on a given revision). It wasn't exactly clear which parts of the demo were the rich client and who knows what performance on that or the thin clients will actually be like. But the demo was certainly a lot better than what I'd previously seen of LWP (ie: a seemingly bloated, slow featureless java framework). Part of that may have been getting an explaination of the target market being the majority of corporate desktop users who are "task workers", not the hard core developer who wants full access to the minutia of their system.

Licensing issues and IP reform seem to be getting more traction too, which is good. Currently it seems like a lot of sniping between the big corporate players about who's behaving worse or better around IP issues. But if the GPL v.3 comes out in the next year I think we'll see more people and companies coming together behind getting things fixed.

The EFF is reporting that the broadcast flag is back from the possibly dead. Word on the street is it will be sneaked into an appropriations bill today. I sent a message to my old California senator. Hopefully since this was on Slashdot the appropriations committee senators will be slashdotted.

NPR affiliate WBUR Boston's "The Connection" aired yesterday a lengthy interview with Phil Torrone and Dale Dougherty of Make and Rosalind Williams from MIT.

It's interesting to hear that beyond encouraging people to get into DIY tinkering in the digital age, the creators are disturbed by consumption and the throw away culture. I hadn't thought of that angle. They definitely provide a forum that will encourage some serious innovation and interesting advances if legislation and the producers (and the negative connotation that the public tends to hear in "hacking") don't manage to squash it.

If nothing else they've encouraged me to fiddle with some things I wouldn't have otherwise.

This is mostly a redux of similar writing by Lessig lately, but it's cool to see it in the Tech Review which probably hits a different set of eyeballs than Slashdot and Groklaw.

There's a rebuttal and a rebuttal to the rebuttal. Epstein maybe does a better job than some market apologists, but really just props up the status quo for no clear reason, especially with his section on buying Microsoft Office for $500 and that being such a compelling benefit to productivity to more than offset the cost.

Personally that is a huge issue to me (and the tangential issue of why people supposedly can't yet productively use OpenOffice, which the press consistently portrays as seriously defficient). I know how to use a word processor. I started on AppleWorks on an Apple ][, moved on over time through WordPerfect on Netware and DOS, various Microsoft, Lotus and now open productivity applications. Aside from embedded images, AppleWorks was capable everything twenty years ago for which I use office apps today. Maybe I had to stick in a separate floppy to launch the spell checker and turn over that floppy for the actual spell check, but still. It worked and well considering the option of writing things by hand or on a typewriter...that difference was the primary difference in productivity.

I'm not sure what that old AppleWorks version would cost today given inflation and maybe in constant currency Microsoft Office is cheaper. But it sure seems like the 99% of features that most users rely on should now be commodity. Is that extra 1% worth the $500? And are those features truly ones that significantly matter to productivity? I say no.

Spending about seven of my first twelve years on the edges of tornado alley (Lubbock, TX and Denver, CO) left me with fascination and awe for tornados. Reading Heavy Weather (and of course the movie Twister a year later) opened my eyes to the interesting technology and science of the storm chasers in the 90's.

The other day I was watching a replay of a Nova episode from last year on the subject. It's interesting to see how the science is progressing.

This week National Geographic has some new video (beware slow loading flash) from a probe that scored a direct hit on an F4. Neat stuff! It'd be cool to be working on tornado probe technology.

It's official, after weeks of rampant rumours! The rumours never really made sense to me, but I guess IBM just couldn't live up to the early promise of the 970 (which included low power potential). I was inclined to think the rumour mill was wrong and Jobs would announce a G5 laptop, but that was just wishful thinking I guess.

Maybe the transition wont be too hard, especially if Apple is focusing on new users and switchers (ie: assuming the iPod halo effect and people's frustration with Windows security continues/worsens). It'll be the hard core Apple fans though that have a large number of harder to port applications.

Beyond the simpler logistics of changing, I wonder about the implications for the future of their OS. Currently one would run a Windows application on a Mac in Windows a complete virtual PC, but I wonder if this will drive things that would allow Windows apps to run more natively. For instance wine becomes an option. At that point the lines separating the user experience of a given OS starts getting blurry and Apple possibly looses its clean UI advantage.

Winn Schwartau is author of Information Warfare, among other books, and is a information security luminary. He's fed up with Windows and has decided to move his company to Macs. He's tracking their move on this blog. There's mention there of the tonnes of wintel issues that have driven him nuts and than his son uses linux, apparently without those problems.

Interestingly, also in the news today is Intel's new CEO saying the best short term solution for home PC security woes is to use Apple systems.

I've gone on the record saying that my PowerBook wasn't all I was expecting, but it is a simple no-hassle box for web surfing, ssh'ing and doing multimedia stuff. Plus the OSX WiFi support is just so simple compared to other OS's. On the multimedia front linux is slowly making progess, but linux wifi's definitely not at the point where I'd give a normal end-user that travels a linux laptop.

It seems like there should be some sort of a way to weight the blogosphere such that "interesting" new things filter to the top. But not all blogs are equally read, nor all readers equal.

In some ways this reminds me of building a personal web of trust with signing private keys in public key encryption, which never really went anywhere. But were this accomplished in blogging it would add to the credibility of news coming out of blogs. Really it all comes back to a topic Lessig covered nicely in Code: Having a way to identify ourselves in cyberspace and to trust the identity of others is really useful. Balancing that with privacy and the desire at times to be able to be anonymous, things generally (barely still?) possible in realspace, is the hard part.

That my employer appears to be asking me to blog externally is interesting, but seems a little risky personally. There may be guidelines, but they're just that...high level guides, which by necessity must leave a lot of room for interpretation and judgement. Guidelines or not, Google search results for "blogger fired" have to give one pause.

So to follow their suggestion: “The postings on this site are my own and don’t necessarily represent IBM’s positions, strategies or opinions.”

Not that any sentient being would have confused that. Maybe I should add a footnote somewhere on the page to reiterate that this page is mine personally?

And for the record, these days I'm working as a filesystem developer focused on IBM's SAN FS and GPFS products on linux.

Following up my post a couple days ago on their new Longhorn advertising comes more insanity. I saw this on Slashdot the other day, but didn't really think about it too much. It's the same old Microsoft, right? But then today I stumbled across this treatment of the issue which mentioned that the rules require submissions be the "sole work and creation of the person submitting the film."

So Microsoft is running advertising that encourages customers' mixing and mashing, but at the same time is running contests which discourage it. Okay...

I knew my wait for HDTV would pay off. I hope Motorola brings these to market soon. I love plasma's colour, but was convinced of their lacking speed by watching a soccer match on one. LCD and projection colour just isn't there for me. Hopefully these come to market and aren't just hype.

The fight against the FCC, major studios and their broadcast flag has resulted in a win for freedom and innovation. I'm not sure if the ruling can or will be appealled, but it's a good start.

Last week I finally managed to find a physical copy of Make. It sounded like a really cool magazine, but I wanted to see it before buying a subscription. It is indeed really cool, although I was hoping it would have content and language that wasn't quite as focused at the adult geek. I know a young inventor who might've been very into the content and subject matter, but I think it's actually a fair amount over her age. We'll see though.

They also have a Make blog. I might have to try the moss grafiti on my rock walls.

The other day I was watching TV and saw a new Microsoft commercial. At first I didn't know what the ad was for, but was curious because it was clearly pushing some sort of technology and they were talking about ripping and mixing and mashing music, among other types of creative production. I was thinking it was strange that a prime time TV ad slot would promote such contentious [sic :)] practices. After a little while the ad turned out to be from Microsoft of all places, touting the features of their three year old Windows XP. They've got a new big ad campaign apparently since Longhorn's so delayed. I wish I could find an online copy of the commercial as I'd like to hear their spiel again and pay closer attention as it seemed so disingenuous in light of all the IP control that Microsoft and other big media companies are pushing (yes I consider Microsoft a content/media company not an irrelevant technology/software company).

Anyway, this BoingBoing post today reminded me of the ad because of it's mix/mash/collage focus.

I've been fiddling for a while with my gallery install trying to get it to run over SSL with mixed luck, but finally got it:

• First my port 443 virtual hosts needed 'SSLEngine on' set and httpd.conf needed a global reference to the cert and key files (even though these can only be set once and are set in the _default_:443 virtualhost).
• Next I needed a 1.5 gallery to have php accessing _SERVER instead of HTTP_SERVER_VARS
• And my gallery config of course needed it's gallery and albums urls set to reflect https:
• An apache RewriteRule forces all accesses to https instead of http.

It is annoying though that gallery's architecture doesn't allow just the login/admin portions of the application to be secured, as I've done easily with MovableType and just a RewriteRule.

Now I just need to find an economical route to getting a server cert that works seemlessly.

This is the sort of thing I've been expecting for years since I really started getting involved with and using open source software and linux.

I've never understood the people claiming that OSS is anti-capitalistic or anti-jobs, because it just stands to reason that with scarce resources cooperation allows you to do thing you couldn't otherwise. And the above article is evidence of that. Aren't the ideas of cooperation and specialisation ones that are taught in ECON 101 as being things that allow advance? It's possible to get places we want faster and speed up the market by working together at it. Personally, I think this is more likely to be the case than the doom and gloom future Microsoft pushes. I see OOS as playing into Ray Kurzweil's belief that technological advance is accelerating and mirroring (from a development perspective) the ideas that IBM and other high end tech vendors are pushing with their "on demand" initiatives.

It's interesting to read that Lessig's thoughts are increasingly about free culture and how non-free technology is hurting the advance of culture. But at just the same old technology and economics level where Bill Gates and his like are arguing freedom is bad, the article has a great quote regarding why Brazil has so gotten into free software and the creative commons...

This rumour would make sense for an Apple Mac mini media center. They could build their own PVR off of mythtv probably that would be more featureful, but Tivo's got a good brand name that they could leverage. With Apple's brand image right now, I'd think an Apple owned Tivo would have better luck with the cable/satellite companies who lately are trying hard to sell their own PVR's.

I wonder if they'd still ship standalone Tivo's with linux as the embedded OS?

I really like my Tivo and it seems like the things possible mythtv are huge...I can imagine a really rich client/server media center in my home that allows viewing and listening from multiple locations in the home (or maybe even outside if I'm on the road and am missing "my shows"). Not that this is anything new. I remember ten years ago reading about Bill Gates fancy new house in Seattle that had technology heading in this direction.

The only hitch is that the recording and motion picture industry heavyweights don't want this to happen and have convinced the government to mandate it doesn't happen via "the broadcast flag". I haven't really had the time to build a mythtv setup in my house, but would like to be able to do something like this in the future. If the media industry has its way though, come July of this year it would be illegal for me to do that (ie: we have to imagine the broadcast flag would nearly always be on and thus recording be prohibited).

Hopefully the courts though are finding legal reason to allow a rich ecosystem of products for home media to develop. Comparing mythtv to Tivo and taking into account the industry's efforts to block online music distribution, I have to believe that open garage-style development is going to create better featured home media products faster than the industry. And that wont be allowed to happen with the broadcast flag.

The local news is reporting that Google's signed an initial deal to buy land for an office in The Dalles (way out in the middle of nowhere). They're talking about some pretty high paying jobs (avg. $60k) and a large number (50-100) for an area that's very rural. I'm curious what they'd be staffing out there.

Not too surprisingly DRM continues to not work well. The boingboing link mentioned in the article is here. I saw it the other day and didn't really give it a second thought (Napster doesn't have any music I want to buy at the moment). Besides...

Copy protection has never worked and is not likely to ever, just based simply on technology facts. Sure they can legistlate and require all sorts of silly hardware and software to get in your way, but where there's a will there's a way. The piraters will always pirate. In the meantime the rest of us are expected to pay more for something we can do less with.

I just want a system that allows me to get music (preferably not one time only on volatile or DRM encumbered / expiring media) and listen to it wherever I am. I used to get that when buying a CD or vinyl recording (and possibly copying it onto some other more portable media like tape or flash memory for running around with), although loosing my "master" "copy" was a bummer (where is my Gorilla's CD). But I guess I'm just a bad consumer for thinking the industry could be using technology to better enable me to consume in that model instead of putting hurdles between buying and listening.

But the media and RIAA presumably will spin this as "3vil h4ckerz" depriving Honest Artists of a living.

Just a few days ago we have him out that Microsoft's latest anti-piracy moves which validate whether the user is on an official windows copy (fair enough) has a validator which specifically targets and disallows Wine users. They may not be on a Microsoft licensed OS, but they're not on a stolen copy either. So much for interoperability. I guess Taylor was just spinning for MS.

Martin Taylor was this week's Slashdot interview victim. He gave a very good interview for the most part. Around TCO though he primarily focuses on business advantage...seems to miss the extreme price advantage and competitive nature of linux on the desktop or in the home for "regular" users, especially ones interested in being able to legally have a full stack of office and security software.

I don't see anything ground breaking in them in as much as they're like Yahoo or Mapquest maps. But they sure have some nice eyecandy (unusual for Google). Still the page layout is much easier than the competition's advertising bloated pages, in Google's typical fashion.

I was hoping to grab one of these last week for my flight to my grandma's funeral, but availability on them is still tight so we just added our names to the waiting list at the Apple store (3-4week wait expected). So I was surprised to get a call yesterday that they were in.

Jenn and I both bought the 1GB model. So far hers caused her windows computer to crash. Installing the software on my Powerbook (iTunes, Quicktime and iPod Installer updates) has crashed the computer and left it in a wierd state.

How hard can it be to make a reliable USB Mass Storage Device? Plugging in into linux shows a simple USB mass storage device with no partition table...presumably I can just write audio files to it and listen to them?

UPDATE: Seems like the iPod firmware updater crashed and somehow hosed my YellowDogLinux installation (which was mounted in OSX read-write). That left the ext driver in OSX trying to fsck and mount the hosed YDL drive. I kind of needed to reinstall YDL anyway after I messed it up a bit upgrading to 4.0, so I did. I also set OSX to mount it read-only normally. At that point things started working again. Jenn had a few more Windows blue screens at the end of trying to transfer tracks to her shuffle, but then it started working.

I don't think this is exactly news, but Linus has been quoted saying patents are a problem. The article says thought that copyright would be a better way to protect software innovations. That seems wrong. In the US copyright is eternal and implicit, so everything written is copyrighted and the copyright lasts forever. Patents have to be applied for (even if most are granted) and expire. Seems like copyright is a worse problem today in America.

The Beeb's recent two part interview of Bill Gates has him pretty much describe that Microsoft and Apple are competing for the same space (without really saying it). Based on his criteria (simple and rich) OSX and a Mac Mini seem positioned to win in the media center market that is Gates' showcase for MS Windows.

And here I thought it was bad that some dumb people write web pages that don't work with non-MSIE web browsers. But who could have known it would get you arrested?

Is this the post-9-11 world we live in thanks to George Bush and John Ashcroft?

As seen on BoingBoing (which has more details than the BBC link above).

It was interesting to see two different posts on Slashdot this last week about the FBI's Carnivore program, since I've just started reading Lawrence Lessig's Code and Other Laws of Cyberspace. So far it's made me realise something I've known for a long time; I need to read the Federalist Papers.

I've always tended to think that things like Carnivore are an unauthorised search and siezure and would be clearly illegal. I'm starting to understand more about why there's ambiguity. And that just motivates me more to understand the framers' intentions.

At any rate, it's largely moot since the private sector is happy to invade our privacy and is sufficiently able to mine much of the data the government wants. So far (I'm only 30 pages in) Code has seemed outdated in a few places, but it is six years old and the internet changes so fast, plus you throw in Sept. 11, that it's not too surprising. But then today Marketplace has a segment on this very sort of surveillance and I'm reminded that the legal issues still aren't clear and if anything the trend is towards more invasive monitoring and an erosion of our rights.

I finally finished moving my various digital images into Gallery. I'm still not convinced it's the software I really want to use.

This sort of thing is great for the conspiracy theorists. It does make you scratch your head and wonder why they'd want so much bandwidth. Maybe they're just trying to save some money. Maybe they get more spam and malicious network activity than most. Maybe they do actually have something fancy up their sleeve.

Makes me wonder further though...I tend to look at the "About" section of companies and their job postings in addition to their products to get a feel for what they're about. It would be pretty easy to trick people by posting bogus jobs though. Come to think of it, Google's already done just that.

The CIA's updated its period forecast (published every 5 years) on the world 15 years in the future. Their thoughts on the year 2020 are interesting. It's a mixed bag. Some really interesting and positive things happening and some scary.

Reading about all the upheaval and change they expect it's fairly amazing that they say they don't expect major conflict (ie: world war), but cite the US's opportunity to guide things for the better. I've worried since Sept. 11, 2001 that the country's foreign policies could end up on a slippery slope dragging the world toward a broad war and Bush has continued to scare me.

Apple appears to have released a really good set of upgrades to its product line this week. The mini iMac is the perfect upgrade for my parents from their 4 year old PC. I haven't used AppleWorks in the Apple ][c days, but iWork sounds like a much needed update (and surely is easier than getting X and OpenOffice to run nicely in OSX until the possible day when there's a native port). And the iShuffle...

I've been wanting a large capacity flash based mp3 player for a while, and the only reason I didn't just get an iPod was I wasn't sure if their harddrives were truly safe to run/bike/snowboard with. Well that and I have a 30GB hard drive Nomad that I bought almost four years ago for the same price as a comparable iPod today (where's Moore's law), but it crashes all the time and has horrible battery life so it was some of the worst money I've ever spent. So the iShuffle has a price point and features I want (solid state, enough capacity, Apple reliability).

I was a little disappointed in the announcement as press I've read made me wonder if a G5 based collaboration with IBM on a media center / gaming console was maybe in the works. But it turns out the mini iMac may be setting the stage for a media center after all. I've also been thinking that they're increasingly in a position to gain some enterprise desktop market share too, which eweek also comments on here...I run a Mac for things at work just fine. I also run linux just fine everywhere.

Still wish Apple'd gone with linux under OSX...

I'm still trying to understand what my own personal thoughts on IP are (the laws are obviously seriously skewed, but is proprietary inherently ok, not-ok, somewhere in between?). Today's announcement by IBM that they'd open a bunch of patented software was picked apart by Information Week as running counter to IBM's presumed need to have strong IP protection, as evidenced by their trying to influence European legislation in that area. I'm not sure there's really a contradiction there. It's a big company so who knows how linked events are.

But as an "outside" observer, IBM is talking a lot about services built on open infrastructure. Even Richard Stallman recently said that software can be proprietary without being antisocial:

Custom software is meant to be used by one client. There's no ethical problem with custom software as long as you're respecting your client's freedom.

So for IBM to talk about open infrastructure and custom solution services and then apparently back that with actions, but still want IP controls for that custom slice of software doesn't seem that surprising or out of line to me.

CD sales keep going up, despite years now of claims from the RIAA, MPAA and misguided artists (ie: LLCoolJ) that the sky is falling. The article notes it's the first increase in sales in four years, but at the same time each year music sales are driving record profits.

Personally I think that says album price increases are a bigger threat to sales than technology.

The labels need to learn to embrace technology instead of concentrating on milking the status quo.

A lot of people are saying lately that they don't see linux rising on the desktop anytime soon, but they are recognising the viability of open source software as evidenced by Firefox usage. The interesting thing is that with apps like Firefox, Thunderbird, and OpenOffice on the rise, Windows users will increasingly find it easier to move as their core productivity apps are portable. The average open source product today is notably portable between linux and other unices plus OSX and increasingly windows.

I first used open source (perl and gcc) in Windows 3.1. As a developer I found the open tools friendlier and then found using them in linux that much friendlier. Now I use a Mac and PC's and run many of the same applications in OSX and Linux.

The ACM Queue has the results of a study into why software projects fail. I'd have thought project complexity and requirements volatility would be some of the most cited reasons. But the top reasons are actually not using proper software engineering methodology, not involving customers and not user proper project management practices. Process and user-centered design make for solid projects! Imagine that...and imagine everybody saying, "But we don't have time for that." Nobody seems to have time to succeed.

Last week a flash movie pseudo-documentary on a gloomy evolution of search technology made the rounds. This week brings some closer speculation on what the future of search might hold.

The NYTimes mentions some of IBM's text analytics research projects in a brief article on searching. Everybody's been waiting for a major breakthrough in artificial intelligence, but what we're really seeing is a slow progression over the years in software's ability to draw correlations. It will be interesting to see if these projects evolve into working products.

MIT's Technology Review (great mag) has an in-depth analysis of the looming war between Microsoft and Google. A few nits with that author's conclusions though...

I wasn't thinking about the fact that SSL encapsulates HTTP. This means that a browser requested site's domain name isn't visible in an incoming SSL connection initialisation and the server side. So the web server can't do name based virtual hosting and have separate SSL certificates per virtual hosts' domain...it wouldn't know which certificate it should have used until after it picked on. So I apparently can only have one of multiple domains served off my box not give a browser domain name mismatch warning.

That's annoying. I wonder why the browser then doesn't do something like get the IP of the domain listed on the server's certificate and compare it with the IP of the domain requested? SSH is okay with multiple keys per server.

Digital cameras appear to be on lots of people's wish lists this Christmas and their broad acceptance in the consumer market is evidenced by the proliferation of sites which allow one to post, share and print their images.

Being a nerd I serve my own stuff up, never quite satisfied with commercial offerings and wanting to have control and learn something too. I've installed and tried linpha and didn't like that it balked at a lot of media types and it's ugly themes (granted now that I have an vague notion of CSS I might've been able to do something about that). I'd heard gallery was better, so today I spent some time dealing with it, but am not impressed with how much of a pain to use it is (although it's much more friendly for an end user I think than linpha), especially from the standpoint of importing an initial mass of somewhat sorted images.

In the end I'm not happy with either. Neither seems that easy to customise and besides I don't want to spend lots and lots of time programming to get something that sorts and orders my images in a flexible way and provides thumbnails. Gallery is very feature rich, but it's not got a good UI for getting at the features easily. LinPHA is simpler, but lacks some features I'm interested in.

I've been following linux kernel development for many years and Jeff V. Merkey's name has occasionally been at the head of the list of annoying people on the linux-kernel mailing list. Recently he popped up offering money for the linux kernel and people flamed away.

He's back and it turns out he was working on behalf of the Cherokee Nation!

Everybody's always arguing about open source code quality. This week Wired's saying linux is better. They quote numbers from static source code analysis. My problem with this is that it just looks at implementation. Bugs can be implementation or design. I think an open, iterative approach which can unapologetically drop old interfaces/implementations can generate better designs over time as well as better implementations. A better design will be cleaner and simpler and a better implementation will be more likely to follow as an implementer can better understand what's going on.

I've worked on lots of products full on implementation bugs that follow directly from poor design and especially design/implementation documentation over time. The linux kernel and many open source projects have an implementation that is hashed out collaboratively online and in a searchable forum, leading to at least some documentation.

This evening's Marketplace broadcast included a blurb about the Firefox release 1.0 advert coming soon in the NYTimes.

It's only been a month and a half since they called for supporters to donate. Hope the ad runs soon.

Everybody's blogged this article this week. It's annoying to read about new technologies and have to wait so long to get them here. Plus in the Bay Area it was always in your face because so many asians would bring things home from travelling and you'd see a decent amount of them, but not be able to buy them (except maybe at a huuuuge import markup).

Reading their reasons though makes me kind of glad our culture and market don't support the sale of the latest and greatest stuff until it's become more commoditised. Although their emphasis on quality infrastructure tendency and towards small and integrated products would be welcome compared with the supersized and redundant wasteful consumption while crippling basic infrastructure with underinvestment that's the normal mode of operation here.

Surely I'm a nerd if one of the best moments in the Baroque Cycle was the experiment Newton does involving sticking a knitting needle in his eye. Not only is it real, but you can see the real notes and diagrams Newton drew during the ordeal.

Found on introvert.net via mindhacks.com.

I'm surprised by the findings of a study that the CSM is reporting:
"...while students seemed to benefit from limited use of computers at school, those who used them several times per week at school saw their academic performance decline significantly as well."

It's certainly no surprise that educators say people need guidance in how to use computers in education. Computers aren't inherently a positive change agent. What I'd like to see is a breakdown of what the students spent their time on while in front of the computer. I'd hazard a guess that kids spend most of their time chatting, playing games, downloading and consuming mass media. The article says, "Computers seem to serve mainly as devices for playing games." But then it does admit that controlled use has been shown to help education.

My mom's mentioned that most students come into the library where she works doing research for schoolwork and are only concerned with the internet. It's definitely a shame if kids are not being encouraged by adults to use a balanced set of resources and aren't cultivating a joy of books and exploration through reading.

Lycos' efforts against spammers have been in the news a bit lately. Suddenly people are crying about it being vigilantism and overzealous. Boo hoo. I wish them luck. An open, distributed response to spammers seems like a good idea to me if it can destroy their business model and discourage people from starting. It would be nice if a system like this established a detente and actually resulted in more available bandwidth instead of just consuming more.

Unfortunately it hasn't put a dent in my spam receipts. I've been getting about 120 a day lately...Almost double what I was getting this summer, with a steady increase inbetween. Thanks to spamassassin I only see a few of those.

I'd not heard of these previously. I wonder how hard they are to get properly alligned and how strong light they need? Clever design though.

I'm not a fan of all the bogus patents which have been being filed. Today Marketplace is reporting that Congress has given the USPTO some headcount to help with the workload. Some argue that the workload is the reason for the bad patents being granted. So maybe this will help. There's just as much to the argument that it's the applicants who are the problem, so more people may just end up meaning more bad patents. We'll see.

I wish there were some more details on today's announcement from IBM and Sony on the "Cell" processors. I'm curious what fundamental changes their designs bring to computer architecture. From what I was able to read today it just sounded like system on a chip.

And the thanks goes to stem cells. The articles I've seen are a bit short on details but I'm sure there will be a lot more press about this.

An outright ban on stem cell research just seems insane. In some ways it feels like the US goverment would like to return to the 1600's and rouse the inquisition.

Your are either with us or against us. And as we all know they have earned capital and now they intend to use it.

The Economist is really becoming a leader in reporting on technology. I've been thinking more lately of getting a subscription, either print or online.

I used to subscribe to c't, published by Heise out of Germany, for my technology news fix. But it got expensive once I wasn't a student and the dollar continued to drop. Plus since they went bi-monthly it's hard to actually keep up with the content.

I have the same problem with the Economist. It often takes me a couple weeks to get through one of their weekly issues. So the dead tree version seems wasteful, but I can read it more often and easier than the online version.

The MIT Technology Review has an article about the Intellectual Property Protection Act under consideration by the Congress. It's yet another in the continuing series of poorly thought out bills around intellectual property control.

And last week's huge federal spending bill included funding for a copyright enforcement czar:

"The legislation, part of the bill funding Justice Department operations, also for the first time funds the National Intellectual Property Law Enforcement Coordination Council (NIPLAC).

NIPLAC is charged with establishing policies, objectives and priorities designed to protect American intellectual property overseas and to coordinate and oversee implementation of intellectual property law enforcement throughout the government. While NIPLAC has been around since the early 1990s, it has never done anything, and appropriators hope that giving the organization $2 million and a new charter will make the office effective."

I wrote an essay ten years ago in college about why there was nothing broken with current copyright law. I'm still not convinced anything is wrong. I should dig that up though and see if my arguments then still hold water with the advent of peer-to-peer sharing, faster internet connections and better compression algorithms.

The Register has a new Nicholas Petreley report on Linux vs. Windows security. It provides a detailed and intelligent analysis of the situation. To somebody familiar with the issues and possessing elementary reasoning faculties (eg: not Steve Ballmer), it doesn't really amount to anything earth shattering. And while it handily debunks Microsoft FUD, I wonder if it does so in too great of detail?

Will the CTO/CIO afraid of open source have their eyes glaze over at this level of detail? It would be interesting to see survey info on executives' technical abilities and levels of intelligence and to know more about how they manage at the intersection of business and technology. But perhaps the persistence of MS Windows is actually tied to factors other than security considerations. Like for instance Microsoft's belligerent monopoly.

The Ansari X-Prize has spawned a new prize. The WTN X Prize is a similar concept, but not specific to private space ventures.

Has anticompetitive behaviour in technology (eg: bogus patents) led us to a point where we need a non-market prize to drive innovation?