EvolvingWorld » HiTech

747-8 near Mr Ranier

tomh — Thu, 11 Feb 2010 15:30:58 +0000

Boeing put out a nice picture of the Boeing 747-8 flying near Mr. Ranier. This such a beautiful airplane, and the perfect setting. The new wings and the longer fueselage give the airplane a new grace. It’s always been big and impressive, but this is definitely the most attractive version. I still remember in 1972 flying to Europe with my parents on a PanAm 747-100, and it was really impressive. It’s just so amazing that after all these years this design is still evolving and staying fresh.

Boeing 747-8 takeoff!

tomh — Tue, 09 Feb 2010 06:04:51 +0000

What a beautiful plane, and check out the takeoff….. it’s big, it’s looong, it’s beautiful and it flies! It’s so cool that 40 years later this is still a current design — the new wings add so much to it. What I find too amusing is the difficulties Boeing had integrating their new composite wings, all designed on the computer, with the fuselage, much of which is still based on old pre-CAD drawings from the late 1960s… As always Flightblogger has the best pictures and the latest Boeing news… go there for more great pictures.

Boeing 787 Dreamliner taxis and rotates!

tomh — Sun, 13 Dec 2009 15:41:01 +0000

Ever since I was a kid I’ve always enjoyed watching airplane takeoffs. The 787 has been of double interest to me in this respect. Technically this is as much a leap forward as the 707 and the 747 were – where they brought routine jet travel and the widebody long distance aircraft to the market, the introduction of the all composite airplane will revolutionize air travel just as much. So it’s with excitement I’ve been following the development of the project, agonized at the delays, and waited for the first flight with bated breath. At the same time, as someone in management for a long time, I can only shake my head at the weird ineptitude of Boeing’s senior management, who seemingly lost in a haze of powerpoint haven’t a clue as to what their engineering teams are up to. At times they’ve give press conferences making announcements saying a milestone was met only to retract them the very next day saying there was a large delay. They seem to have mostly lived in a world of wishful thinking, only to rudely awakened to the fact that their project is using a lot of brand new technology which is somewhat unpredictable, being brand new technology. This is on top of the disastrous attempt to get the suppliers to to do all of their integration work for them, which they might have guessed would have been problematic especially given the radical nature of the technology on the project. But after all that, the engineers produced a beautiful airplane, and it looks like it’s about to fly in the next few days, weather permitting!

Jon Ostrower’s nice FlightBlogger blog is carrying lots of coverage of the airplane. Here’s some video of it on its taxi runs (with rotation!) yesterday:

Videos Courtesy of Liz Matzelle

Synology,LLink and Pocorn-hour

tomh — Mon, 12 Oct 2009 14:13:08 +0000

Much to my wife’s dismay I have an enormous collection of recorded live music, which occupies an entire closet, one my wife wishes she could reclaim. As my collection has gotten larger two issues have come into focus. (1) The continuing change in formats means that a good bit of the collection is always in danger of obsolescence, and (2) it’s cumbersome to locate recordings. I’ve resorted to tagging things with numbers so I can look them up, much like in a library. So the idea of storing music on a hard disk and and simply browsing for it has long held appeal. One challenge is that my collection does not fit the usual itunes paradigm, in that I don’t have songs and playlists. Each concert is an integral whole, and is most easily represented as a directory containing an ordered sequence of flac or shn files. So any solution that is just a bag of songs is useless to me, I need connected series of songs.

Last year I prototyped a solution. I got a popcorn hour A-110, which is basically a network-to-hdmi bridge. It lets you stream video, pictures or audio from a network connection to yours stereo. This lovely device presented the problem of how to get the network to it. The stereo is on a different floor than the wired network, so I would need a wireless network. I put in a dedicated 80211N link to avoid contention with our normal wireless network. Particularly video stresses the network to the maximum. Obviously I can’t stream Blueray this way, but fortunately that’s not one of my requirements. The next question was how to organize and transmit the music. I opted for llink which lets you organize your media in a directory tree, then lets you browse it with via web pages and transmits it via http. This solves a couple of problems. http is by far the fastest protocol to devices like the popcorn hour. The hierarchical presentation is ideal for my kind of collection. For now I simply put the files on my pc.

This first solution worked ok. It did exhibit a number of problems. First of all, for unknown reasons, the popcorn hour often couldn’t see the PC, forcing me to restart the llink service and/or reboot the A-110. Second, it wasn’t real fast, with considerable gaps between songs, which is a problem if the music is continuous. Third, I was running out of disk space. And finally, most critically, I was depending on a single disk drive, which could fail at any time, with no redundancy or obvious provision for backup.

After having studied the matter for a long time, I finally bought a Synology DS-509 NAS. I populated it with 5 1TB Samsung Raid-class drives. I decided on a Raid-6 setup. Raid-6 degrades write performance, but my filesystem is mostly read-only. Conversely, if one disk does fail, the rebuild can take almost 18 hours at this size of file system, and the stress from that rebuild could well push another disk over the edge, so being able to recover from two failed disks is important. So now I have a 3TB of disk space, in a tiny, compact, silent and cool box. LLink installed very easily on it (fortunately you can ssh into the DS-509, it’s just linux).

So far I’m very pleased with this setup. It’s scalable, fast, and reliable. And if I manage to run out of disk space, I can add an expansion cabinet. The web based interface makes it easy to administer. It’s much faster than my old setup, and is always instantly recognized by the A-110. A lot of the quirky behavior I attributed to the A-110 seems to have just been interactions with my Windows PC. The A-110 has been much better behaved since it’s connected the DS-509. Interestingly, when copying files to it, I often wind up using ftp instead of the explorer, because ftp is so much faster.

Next steps are upgrading my wired network to Gigabit ethernet, chiefly by replacing the routers. This will let me stream video more effectively, as right now the wireless links are faster than the wired connections! I’ve ordered an C-200 from popcorn hour, a device which I find fascinating if they can only make it work, but that’s a story for another day.

Fun with browser compatibility

tomh — Wed, 16 Sep 2009 16:35:44 +0000

Ever since the divergence between Netscape and Internet Explorer browser compatibility has been a major challenge for web designers. At Art in Action I’ve been very concerned about this, since our users are often very technically unsophisticated and we can presume run wide range of browsers on sometimes fairly old hardware. Proof of this comes when we ask some users about version of the browser they’re using, and they respond with: “What’s a browser?” .

As I’ve dug into things though I’ve discovered that this is one problem that’s actually becoming fairly manageable. Simply, the newer browsers are better behaved. If you stick to very pure, syntactically correct html and javascript, you actually get very predictable results on the modern browsers. (I should point out, as I described in earlier entries, that I rely a lot on portable html design patterns and deep syntax checking of the html to make sure I have a good starting point.) What about the older browsers? Not wanting to setup lots of machines, I found two useful tricks. First, Adobe Browser Lab is an excellent tool for comparing web pages in different browsers. You simply register on their site, indicate which browser versions you want to try, give it a url, and off you go. This confirmed that actually most of the browsers were producing perfectly equivalent results. I was still concerned about IE 6 which is the oldest and worst of the browsers still commonly in use. I wanted to test it more thoroughly. I then remembered my Microsoft Virtual PC instance. When I got my new computer it has Vista on it, and discovered that unlike with XP I couldn’t run a lot of old software anymore. So I installed Virtual PC with a Windows 98 instance. That runs old stuff like the wonderful complete National Geographic collection perfectly. This was running IE 5, which was current back when Windows 98 came out. I then realized that early versions of IE 6 still supported Windows 98, and upgraded the instance to IE 6. Now I could try out the site using a really old browser and OS combination. Interestingly, it all worked fine, which confirms my belief about the value of writing very clean, portable html. Just for fun, I did try IE 5 too, but there the lack of modern Javascript did things in.

That said there are still subtle rendering differences between browsers which will certainly drive a true artist nuts. However, if you are a pragmatist like me, you’ll find that by being disciplined in your work and being a little flexible as to what the precise outcome is, it’s actually not too hard nowadays to produce very portable applications. I will also say that this discipline is greatly eased if you are not producing ad hoc html, but generating it via an engine according to fixed rules. The rigourous use of stylesheets and the fact that in our application almost everything is generated from a higher level representation makes all of this vastly more tractable.

Snow Leopard is here!

tomh — Wed, 02 Sep 2009 05:36:52 +0000

So I got snow leopard today and upgraded my computer. Actually, the whole install went very smoothly, which I think was one of the design goals. Mostly everything just worked. I installed the OS and xcode. Most things simply worked. I had had the foresight to upgrade the Cisco VPN to the latest version BEFORE I did this, since it usually breaks on an OS upgrade. The first real challenge came with the printer, an Epson 600 Workforce. It uses ppc code, and rosetta, the emulation layer, is no longer installed by default. So I had to install that, and then reinstall the printer driver, having updated it from Epson’s site. It now works perfectly. The next challenge came with the unix applications. I had to forcibly reinstall and rebuild everything in macports, but that went ok. Where I got in trouble is with Wine, which depends on a version of Freetype that’s being only built for 64 bits when it’s needed in 32bits. What’s going on of course is standard Unix dependency hell triggered by the migration of much of OS X to 64 bits in Snow Leopard. Well since others have already logged the bug with macports, I’m sure it will be gone in a few days.
So now that I did, do I like it ? Visually the most obvious change is conforming the default gamma setting to the same value as in Windows. The colors on my desktop are noticeably different. There’s lots of other little changes too. John Siracusa’s review on Ars Technica gives an in-depth look at what’s new. The new finder and dock/expose features are nice, it feelds speedier, and lots of other nice little things. It’s not a revolution, but a good incremental improvement. And at $29 the price was right.

Art in Action website is done!

tomh — Tue, 25 Aug 2009 14:38:58 +0000

We did it! It’s always a great feeling to launch a new product. The new Art in Action online program has been a very interesting project, not only allowing me to try out different technologies like Symfony, but for the interesting management challenges the project has posed. In particular, using our team of high school interns (all girls) was quite the challenge. It turned out great and they were both smart and productive, and a lot of fun to work with. But it required a lot of preparation on my part to create appropriate tools so that they could be productive.

It’s also kind of a unique system. Although Art in Action exists to teach students art, this project is not about directly having students go online, rather it’s focus is on teaching the parents and teachers who deliver the curriculum how to teach art themselves. It’s also an interesting mixture of art history and art application. The common hierarchy of concepts in terms of which all the art works are explained is actually a fairly interesting concept from a theory-of-art perspective. It provides a common, reusable framework in terms of which a broad range of works from many different times and cultures can be examined.

Of course, like all 1.0 releases, we have a long list of stuff we didn’t get into the first version, but that’s the fun of software projects. We’ll fill more of this in coming months, and we do have a lot of very interesting ideas.

GM on electric cars

tomh — Sun, 23 Aug 2009 03:48:28 +0000

The other days I heard an interesting lecture at Xerox Parc by Dr. John Hsu of GM’s Advanced Programs office in Palo Alto on GM’s ideas about electric cars. He presented quite a bit of interesting material, a lot of which I hadn’t heard presented before in this form. First of all he distinguished between different kinds of hybrid cars. First, there is the conventional hybrid, like the Prius, where both motors are used together most of the the time and there is very limited electric only range. He contrasts this with GM’s ideas about extended range hybrid vehicles. In these cars a reasonable driving range is achievable on electric power only, and they are assumed to be pluggable, i.e. you can recharge the battery by plugging them in, in addition to turning on the gasoline engine. He distinguished between two types within this category. First, is the urban capable vehicle. This has sufficient range to cover every day driving without recharging the battery, but still relies on kicking in the gasoline engine to achieve peak performance. The second category is the “extented range electric vehicle” which achieves its full performance on the battery and only turns on gasoline engine at all if the battery gets discharged. The rationale behind these categories is that they use progressively less gasoline than a conventional hybrid. They do in this large part by having a bigger battery than a conventional hybrid and being pluggable. The “extended range” category is exemplified by the Volt. The urban capable hybrid by another model which GM plans to introduce (this was to have been a Saturn Vue derivative, but the sale of Saturn has scuppered that plan.). GM has both categories, because the urban capable designs currently allow a broader range of potentially bigger platforms than the Volt design does. He then went on to show a most interesting analysis, as to whether the pluggable hybrids actually reduce CO2 in comparison to regular hybrid. It turns out that the answer depends on the characteristics of the local grid. In California where a lot of electric power is clean, the pluggable hybrids significantly reduce CO2. On the east coast, where a lot of electricity comes from coal, in fact they increase CO2 consumption. Another chart showed the actual fuel savings. Basically the VOLT saves the average driver about half a gallon of fuel per day over, say, a Prius.

So this was a lot of good data, but what does it all mean? First of all, the economics of these designs are still a little odd. A 1/2 gallon a day will not compensate for the sticker shock of the Volt. Particularly as you are getting a glorified commuter car. This will not compete with the Japanese designs. Honda has intentionally priced the Insight to be affordable. Nissan is coming out with an economy priced pure electric car. And remember, today’s Prius is actually quite a large car, good for long trips and carrying lots of cargo. The Volt is not a replacement for that, and much pricier.

An interesting question is the relative sizing of the drive trains. Dr. Hsu pointed this out, that one option is to simply have redundant drive trains, with either electric of gasoline motors having equal power. Obviously, this is very inefficient. The compromise is either to make one motor much smaller than the other, or simply make them both small. The Toyota hybrids go for the later option – and simply use all the motors together for peak acceleration. This means thought that even with a bigger battery, performance is limited without turning on the gasoline engine. Admittedly, with a bigger battery, you won’t have to run the engine as much to recharge the battery, and can only run when you need to accelerate more. So making a Prius pluggable will make it get much higher gas mileage, but without retuning the propulsion system, to drive a significant distance at 70mph will require some gasoline. The Volt’s option is to allow you to go full tilt on the electric motor, but then suffer a big performance penalty when you’re out of battery. This is much what happens in the Prius if you drive at freeway speeds in the mountains – once you kill the battery your acceleration drops way off. With the Prius one learns to take advantage of momentum and give the battery enough time to recharge either from the motor or from a descent so that you do have enough power for the up hill sections. Driving to Yosemite or up I-80 in Sierras gives one lots of practice with that. The question will be if similar behavior will make the Volt palatable too at the outer limits of its range.

Some people wondered why bother with the gasoline motor at all if 80% of all trips are short range, and the Volt can handle that range? The answer of course is that people do need to take the other 20% of trips, and a range constricted car will force people into having a more conventional second vehicle, which dilutes its impact per household. People also asked about ideas for exchanging batteries. Dr Hsu responded that it’s like with changing batteries in latops: it used to be much more common than it is today, few people are doing it, and increasingly there is simply enough battery life not to do that. So from his perspective by the time such a battery swapping scheme came to market, it would likely be obsolete.

So what’s it all mean? I know I want to get a pluggable car for my next vehicle. It’s fairly obvious this will by a hybrid, with a gasoline engine in it, rather than a pure electric. I demand a practical car that can do I anything I want. It will also need to be affordable – I won’t pay an absurd premium simply for unusual technology. So, what will it be? Will I buy one of the GM electric cars? Honestly, at this point I don’t know. I don’t really see a convincing feature set there at an appropriate price. Of course, with all the new players entering the market practically every month (the latest being Nissan), I’m sure this will all change over the next two years. In any case this was a good, informative talk, even though I still haven’t figured out which electric car to get.

Too many plugins

tomh — Thu, 13 Aug 2009 00:14:12 +0000

One of the strengths of the symfony application framework are all the useful plugins you can get to extend it. I’ve used quite a few on the Art in Action project.
However, as I observe the symfony web site from day to day I notice an interesting phenomenon. Developers love creating plugins. Almost every day there are more plugins. Yet few of these plugins are very useful. Often times the same topic gets reimplemented over and over again. Furthermore, the developers have an early fit of enthusiasm and release quite a few versions, then one never hears from them again. What’s going on here is that people enjoy showing off their code by sharing it, but aren’t really interested in putting in the energy to make it truly general or supportable. Even more so, they aren’t interested in understanding other peoples’ code and figuring out how to generalize it. For example one of the more popular plugins is the very useful Lucene Plugin. In the source repository it already has a branch supporting solr. However, now someone just released a new plugin to do Solr. Why? Why not just fix the Lucene Plugin ? Of course, users aren’t fooled, and in fact what you see is that a few plugins are very popular, but over half the plugins have no more than 1 user, including the author themselves.

I’m very curious now that the Symfony 1.3 release date is set, and especially with the decision to make Doctrine the default ORM, how long it will take for the popular plugins (where we’ll define popular as having > 10 registered users on the website) to get updated. I strongly suspect I will wind up having to port a bunch of them myself. It’s this bit of forthcoming work that’s motivating my diatribe above. If folks wouldn’t do quite as much duplicative work, and focus a little more on building out what’s already released, it would benefit everyone. Of course, human nature gets in the way here. What I’m asking for is deeper thinking and extra work in the short term for a long term benefit, and as many folks are working on projects with very short term deadlines, that short term effort just isn’t practical.

In looking closely at the user group, I observe another sociological phenomenon at work. A lot of Symfony users are consultants who build a lot of sites for clients. As such, they are attracted to Symfony because it’s power allows them to complete their projects faster with much less effort. However, being consultants, they tend to finish one project and then move on to the next – I suspect many of them never maintain their projects much. The result of that is a focus on getting projects done quickly today, but little concern, about say, updating a project from one version of Symfony to the next. Instead, the developer will simply switch to a newer version upon starting some new project. That behavior also reduces the desire to contribute to real reuse. Of course, these same developers benefit greatly when there are truly reusable pieces out there.

All of this shows that the varied motivations of human beings greatly complicate technical decisions in ways not predictable from the technology alone.

Most useful HTML Tool

tomh — Mon, 10 Aug 2009 02:34:57 +0000

I’ve found that for doing web development of any kind an absolutely indispensable tool is CSE HTML Validator Pro. This is an amazingly versatile tool. Basically, you take the source for your web pages, run it through the tool, and it tells you what’s wrong in immense detail. It does this page by page, or in batch mode. It can operate off of live web pages, read files, or by pasting in source code. It tells you about standards compliance, syntax errors, spelling errors, it checks your JavaScript, it checks accessibility, and makes sure all your links work. It’s enormously configurable, allowing you to create custom validation schemes, selectively turn on or off different tests, either as a matter of policy, or on a one-off basis. It works equally well with html and xhtml. You can also directly edit files in the system, which lets you dynamically debug your way through problems. I’ve been using this tool for many years, and watched it get steadily better through many upgrades. I find it useful in a variety of different ways. It’s good for debugging new web pages, particularly to ensure consistent cross-browser and cross-platform operation. It’s good for cleaning up finished pages to check for typos. It’s also very useful to do maintenance on existing pages to check that external links still work. I find whenever I run it against some project I almost always learn something new.

On my latest project I found an interesting way to simplify testing the web site. Since there are many, many pages, with thousands of links, running the pages one at a time through the tool is cumbersome. So I created an automated test that uses CSE Validator. I first created two specialized web pages. One dumps out all the links in the system, and next to each link is a second link back to the spot where the original link was found. The second concatenates all the static html from the site into one long page, again with cross-links back to where the original html came from. Then I connect these pages to CSE, and let it validate them. This way only two pages get executed, but all the static html and links are tested at once! As errors are found, they are easy to fix, because I can use the embedded cross-links to take me back to exactly the spot where the material came from, making it easy to fix the error. In this fashion a ten minute test can verify thousands of pages and external links and make sure all links work, html is correct, and that there are no spelling mistakes.

CSE HTML Validator – never leave home without it!