Parallel programming with ParaSail

An interesting ACM news item recently is a write-up about a language called ParaSail, which is specifically designed for multi-core optimization. It uses a technique called pico-threading, effectively splitting a program into “thousands of smaller tasks,” which can be executed concurrently on however many processor cores the target system has.

Although Duncan Graham-Rowe’s article on the language likens ParaSail to C# and Java, the syntax only very vaguely resembles those modern programming languages, instead taking after languages like Basic or possibly Lua. The syntax is a little ugly to me personally, but I will definitely be watching this project with interest. The implications for high-performance applications are immense, especially with the rising numbers of cores in today’s processors.

On an unrelated note, I just got back from a family trip to Disneyland, which was quite fun. I’ve been very interested in the special effects in Disney’s rides for several years now, so I might write a bit about that later on…

On the Road to V8

That’s V8 the ECMAScript engine, by the way. I’ve been looking at it recently because I’m suddenly feeling very interested in embedding a scripting language in a C++ application. Since I’m out of school for the moment, I have some time to fiddle with open source projects and see if I can get them to work.

I find V8 very interesting. It’s the engine used for JavaScript in Google Chrome, and it’s designed for high performance. Unfortunately, the build process is a little… I don’t know… “wonky” might be a good word for it. Instead of a standard tool that just about every developer has (like make), Google uses an esoteric bit of engineering called SCons for the build process. It’s been updated relatively recently (within the last year), but the “installer” looks like something from the 1990′s with its full-screen background window. Finishing up the installation, because the program available for download just copies a bunch of files into your Python directory, is even more of a hassle than having to look at the so-called installer.

After about half an hour of increasingly despondent searching for answers, I found an article on Google Code explaining where the files were actually installed, and that the installer didn’t put that directory in the system path. Then I realized that the page was linked to from the main V8 build guide, and felt a little silly for missing it. Once that directory was in my PATH, SCons worked well enough, building the DLL and static library with the Microsoft Visual C++ compiler. I have since found out how to compile it with MinGW instead, and although it’s a little more involved, it also works fairly well.

Once I got past the build process, I was quite impressed with how easy it is to use V8. A really basic embedded interpreter can be written in under 50 lines of code. At the same time, if you want to crank up the complexity, V8 is fully capable of delivering much, much more. I’m currently working on a scripting environment based on V8 which could end up being fairly useful; one of the things I’ve already added rudimentary support for is loading native code libraries at runtime. The exports of those libraries are then available to the currently executing script.

So, pretty neat stuff… I’m also hoping to incorporate V8 into a game at some point. The future can be an exciting thing!

Python’s style guide: the good, the bad, and the ugly

I’ve been looking recently at making some contributions to a project that uses python for computational mathematics. This, of course, meant that I had to look up some tutorials on python! Having never done much with the language before, it should be an educational experience. One of the points the official tutorial brings up is the most popular style guide for python projects. There’s plenty there that I think is totally awesome, but there are also some guidelines that I hope I’ll never stoop low enough to follow.

The good: Extraneous whitespace is the biggie here. It’s a huge pet peeve of mine when people think they have to drown every set of parentheses in spaces, and this style guide recommends against it. Bravo. Comments should also be complete sentences — I completely agree with that. Incoherent babbling with an octothorpe at the beginning never did anyone any good. Limiting line length is also good, although it seems to be pretty standard practice. It’s good to explicitly state it though. Finally, the guideline about putting separate statements on separate lines is a good one, but again, pretty much standard practice.

The bad: In comments, this guide says to use two spaces after a period that ends a sentence. There are so many reasons why not to do this. Admittedly, programming editors most often use a monospace font, so this might be a bit more of a gray area, but I find single-spaced text to be easier to read even in a monospace font anyway. Regardless, it’s a bad habit to get into, especially in a field like Computer Science where modern technology and trends matter more.

The ugly: This is a matter of personal preference, but I disagree with the recommendation to use tabs, rather than spaces, for indentation. Spaces force one way of looking at code on anybody who reads it, as well as being extremely easy to do wrong. Having a space too many or too few on a line is very common in code I’ve seen that uses spaces for indentation. The “tabs vs. spaces” argument is one for another day, but I felt I had to mention it here because of its (biased) inclusion in the style guide.

To summarize, the official style guide that the developers of python recommend has a lot of very good points, but falls short in a few crucial areas. While many of the practices they recommend are verifiably sound, there are other points that should be rethought, or at the very least, the reasoning behind them should be clarified.

Adobe Reader auto-launching

After “upgrading” to the latest version of adobe reader, I found myself experiencing an old annoyance again. Long story short, reader would launch two copies of itself whenever I hovered my mouse cursor over, or selected, a PDF file in windows explorer. It does this to display metadata such as author and title in the tooltips for PDFs, as well as providing a thumbnail preview when a PDF is selected. However, I never really use the extra information it throws at me, and the frequent flickering of the mouse cursor (not to mention the sudden spikes in CPU usage) was seriously getting on my nerves.

In adobe reader 9, it was easy enough to turn off this “feature” simply by deleting a registry key that was fairly easy to find. With adobe reader 10 now installed, that fix didn’t have any effect, even after forcing the windows shell to restart. Something a little more elegant was needed. I found what I was looking for, disturbingly enough, at the US-CERT website in an article about a buffer overflow vulnerability. On the bright side, that means that getting rid of this annoyance makes my computer more secure! All I had to do was issue one command at the command prompt:

regsvr32 /u “C:\Program Files\Common Files\Adobe\Acrobat\ActiveX\pdfshell.dll”

Although the article is a couple of years old, it looks like adobe hasn’t changed its overall approach much. Unregistering that one DLL made a world of difference: my PDF tooltips are once again fast and lightweight. It would be nice if adobe’s metadata display system were less clunky. Failing that, they could at least include an option in the preferences dialog allowing users to turn it off.

Firefox 4: Mostly Harmless

Another semester over. Finally. And now that Internet research is no longer so crucial for me, I decided to bite the bullet and try Firefox 4. It’s been nagging me to update for a while, but after seeing the extremely lackluster beta Mozilla had out a while ago, I wasn’t sure I wanted to stake my ability to access the web on the conjecture that they had managed to improve it since then. When I did try the Firefox 4 beta, it was so bad that I uninstalled it the same day I installed it. I didn’t want to have a similar experience while school was in session and end up having to switch to Opera or something.

So, the browser. Overall, I guess my impression of it is positive enough. It is noticeably faster than 3.6.x, which is a big plus in my book. Page loads and startup are a bit faster, navigating back is way faster (not so much need for tabbed browsing anymore), and working with tabs is a lot snappier than before. The tab close/open animations are kind of neat too, although I can’t help but wonder if it would be a little bit faster without those. I’ll have to look into that. The new interface look is not entirely unappealing either, so I don’t feel an overwhelming urge to go get the Firefox 3 theme and restore the interface to exactly how it was. It isn’t that bad.

I was also impressed by the retention of old settings and interfaces. I did my upgrade as a browser update, not by downloading and running an installer, so a lot of the interface stayed the same. I think I avoided the worst of the interface changes because of this, so this review may end up being more favorable than others. The menu bar was still there, I still had my bookmark toolbar, and the tab bar was in the right place. There were a few things that didn’t stay the same, but I found workarounds for the important ones.

Now for the not-so-good parts that still crept through. Combining the refresh and stop buttons was an interesting idea, but it doesn’t give you that much extra room to work with unless you still have a VGA monitor (in which case Firefox might not even run on your computer!). Putting the combined button on the other side of the address bar was downright stupid. I’m sorry, but yes, that was a terrible decision. These were some of the first things I fixed, and here’s how I did it:

1. Right-click on a toolbar and select Customize…
2. Drag the refresh and stop buttons wherever you want them (they separate when you go into customize mode).
3. Put a separator between the two. Now the buttons will stay apart.
4. Click the Done button.
5. Enjoy!

Another major, major problem was the removal of the status bar. Yes, you read that correctly. The status bar. Gone. Instead, there’s an annoying little popup with poor contrast to show you where links will take you, and you have an “add-on bar” where the status bar used to be. The add-on bar looks like a status bar, only it doesn’t work like one. Oh yes, and the add-on buttons don’t work, you have to click on the text descriptions next to them. Fortunately, there’s a fix for this horrendous oversight as well: the Status-4-Evar add-on will bring back the status bar, link URLs, add-on buttons and all! My hat goes off to the developer, Sparky Bluefang. Well done. Well done indeed.

Summary: Firefox 4 is good enough. There are some very serious issues with the new interface, but it’s customizable enough (as always) to work around those problems and make it fairly close to that great old interface we’re familiar with. The speed improvements are very welcome, as Firefox 3.6 was getting a bit sluggish for my tastes. In the end, the real credit for Firefox’s continued usability goes to its customizability more than anything else, and mostly to the add-on developers. On a five-star scale, I would give the browser itself 3.5 stars. Usable with some tweaks, and slightly improved overall from the previous version.

Finally, some progress

Wow. Another semester of college finally over, and I’d have to say it was my busiest yet. Of course, spring ’11 is already looking like it will be even busier, so I’d better write a post or two now while I can…

Things have come quite a ways in the couple of months since my last post. AT&T has actually released a documentary on the dangers of texting and driving, urging people to wait for a less laceratingly stupid moment to broadcast their lives. For all you skeptics out there, yes, a phone company really does seem to care. I have proof:

I like this new direction that AT&T is taking. Of course it’s not nearly enough, but it is a nice little first step in the right direction. The key, in my not-so-humble opinion, is still enforcement. Take away the possibility of this distraction happening at all, and the problem is solved. Otherwise there will almost inevitably be someone who just has to tap out a last deadly “LOL” to one of their friends, regardless of how much blood is spilled as a result.

So yes, kudos to AT&T for doing this documentary. It’s really great and all, but it doesn’t really solve any problems on a significant scale. Now that they’ve shown a desire to do something about texting while driving, however, why don’t they actually do something about it? If they’re having trouble coming up with ideas, may I suggest this? Excerpt:

The ideal option would be to disable any non-emergency calls to or from any phone that’s moving too fast. This could be done at the network level, with triangulation technology that’s almost certainly already in place and being actively used. People could still call 911 while driving, but they wouldn’t be able to send or receive any other calls or messages until they pulled over and stopped.

Admittedly, it would take wider adoption than just one provider to make it effective. A federal regulation requiring the system to be in place by a certain date would probably be the way to handle it, and what better proponents than the very corporations at the heart of the problem right now? Apparently Sprint has been involved in some initiatives as well, so maybe it’s time for a joint request from Sprint and AT&T (along with any others willing to join in) for future legislation to require use of this sort of technology (which they could develop) by all service providers.

What do you think, AT&T? Ready to put your money where your mouth is?

Draconian or Reasonable?

Okay, seeing as this is one of my favorite topics, I have to take a few minutes to write about this article. Summary: is a nationwide ban on all cell phone use while driving “draconian”? We don’t know, and we’re not sure if it can even be done.

First issue: no, it’s not “draconian” or even excessive. It’s actually necessary, as long as it can be enforced (I’ll address that point in a minute). Just about everybody (except possibly military organizations) got along just fine driving their cars without any form of mobile communication for over 80 years, and now suddenly everyone has to be glued to their crackberries 24/7? Even if it endangers millions of lives daily? I don’t think so.

Second issue: yes, it can be enforced, and it’s not that hard. There are several options here, with varying levels of expense depending on whether the communication industry can be subdued enough. The ideal option would be to disable any non-emergency calls to or from any phone that’s moving too fast. This could be done at the network level, with triangulation technology that’s almost certainly already in place and being actively used. People could still call 911 while driving, but they wouldn’t be able to send or receive any other calls or messages until they pulled over and stopped. It sounds like the perfect solution, right? The only problem is that it would have to involve cooperation from the phone companies.

Ah yes, the phone companies. That huge, booming industry, willing to do just about anything to maintain its profit margins. Are they likely to cooperate with anyone if it will make their services less accessible? Never mind human lives, there’s money to made! They can afford to lose some customers here and there in nasty, fiery car wrecks, as long as most of them keep up with their service plans and phone bills. It all comes down to profits, and with somewhere in the neighborhood of 5 billion subscribers, they can afford to lose millions of customers a year. That’s the kind of statistic they would write off as “a drop in the bucket.”

So what other solutions are there? Signal jammers could be required by law, either in vehicles or along major roads. However, this could be prohibitively expensive, not to mention impractical in some cases. Putting jammers in people’s cars would be the most economical method, since they wouldn’t have to be very powerful. A range of 10 feet would be fine, and they could even be tied to the cars’ internal computers, allowing cell phone use if the engine is turned off. Car-based jammers could be disabled relatively easily, though. A black market for jammer-hacking would probably pop up within months of the legislation taking effect, followed by a rash of car “accidents” as people resumed (illegally) texting/phoning while driving.

Roadside jammers would hardly be any better. The cost would be huge, since the devices would have to cover much more territory, and there would surely be large areas without jammer coverage. There would also be problems with jamming signal “spillage,” affecting residential areas near major thoroughfares. While it would be more difficult for phone addicts to disable them, especially with surveillance and heavy protective casings, stationary jammers would not be very effective at solving the problem and would cause plenty of new problems of their own. If I were to rank the solutions I’ve come up with so far, this would be the worst; a last resort only, for sure.

In an ideal world, we could simply educate people about the dangers of irresponsible cell phone use. The same strategy doesn’t seem to have worked terribly well for cigarettes (another major industry eager to protect its profit margins), so I’m very pessimistic about that possibility. That’s not to say it isn’t a good idea to try, but education alone won’t be enough. Bans, and more importantly, enforcement, are the only viable way I can see to effectively stop people from texting themselves, and others, to death.

 

N.B. I’ve noticed a common argument against bans is that they haven’t lowered accident rates yet, and in some cases actually seem to have raised accident rates. This isn’t because cell phones don’t distract people; it’s because they are more distracting when people resort to hiding their phones in order to skirt the ban. Drug addicts don’t stop getting high just because it’s illegal, they just hide the drugs when they think someone’s looking. Enforcement is the key, because many people will not stop doing something destructive or dangerous unless they have no choice in the matter. Sure, personal freedom is important, but so is public safety.

2D Acceleration in Ubuntu

Hi there, sports fans! Yes, I know it’s been a while since my last post (noticing a pattern there?). Again, school has taken over, although at the moment I don’t have quite as much homework thanks to a couple of recent midterms.

Exciting news though: I finally installed Ubuntu for real a couple of weeks before Fall semester started, and I’ve been using it like crazy for all sorts of things! I’m actually quite enjoying having a different OS around for a change, and it helps with school since a lot of my computer science stuff is much easier to do in Linux. I have had a few problems adjusting to it, although I’ve found solutions to most of those problems.

A perfect example is the one I just solved today. For a while now, I’ve been noticing that Ubuntu didn’t seem to have very good 2D video acceleration. Scrolling, especially on web pages, was slow and jerky, and switching tabs in Firefox could take a few seconds on a bad day. That’s with fairly decent hardware, too. Then, just a few minutes ago, I tried updating my nVidia graphics drivers, restarted, and bingo. Scrolling is nice and smooth, tab switching is nice and snappy, and even the system menus seem a bit faster.

It looks like the problem was an old, outdated driver. For whatever reason, Ubuntu had defaulted to version 173; after the update it looks like it’s version 195. Apparently version 260 is out now, but I’m happy with what I’ve got for the moment. My current driver has been tested by the Ubuntu developers, so it’s known to work well, I guess.

In case you’re having the same problem and need to solve it, here’s how:

1. Open your list of hardware drivers (System menu → Administration → Hardware Drivers).
2. Wait for it to finish scanning for drivers.
3. Look at the list once it comes up. If there’s a video driver listed as “version current” and it’s not the one you’re using, select it and click the “activate” button. There will be a green circle next to the driver you’re using. If you’re using the latest driver your OS can find, try downloading a new one from the company that made your graphics card chipset.
4. Restart your computer (and boot into Linux again if you have a dual-boot setup like me).
5. Browse the web with lightning speed! (or whatever floats your boat).

So that’s my tip of the day on how to tweak Ubuntu. Stay tuned for more… eventually… homework and exams permitting…

No, it’s not “just the ‘internet’”

I found this short article on the Wired website a while ago, but didn’t get around to writing about it until now. Summary: Tony Long (the article’s author) doesn’t want to have to hold down the shift key for a fraction of a second when he’s writing about the Internet (which has got to be pretty often), so the whole of Wired News has to stop capitalizing that ‘I’ at the beginning of the word.

Mr. Long likens the capitalization of “Internet” to some sort of attention-grabbing religious fervor, and claims that “it makes no sense in English.”

alright then. while we’re at it, i’ll call the magazine “wired” (lowercase ‘w’), i’ll keep speaking “english,” and i’ll start planning to apply to graduate schools in “california.” let’s stop using uppercase letters at all! and let’s get rid of all that unnecessary punctuation too. lets all jst use txt spk 4 evrythng!

Pretty jarring on the eyes, no? Let’s examine the reasoning behind the capital ‘I’ and see if we can’t make sense of it. This is something I learned in my basic computer networks college class when I was seventeen, and Wikipedia seems to agree with what I remember (links: Internet, Internet capitalization conventions). Basically, “internet” (lower case, generic noun) is short for “internetwork”; in other words, all it means is two or more networks connected to each other. Capitalized, the word “Internet” becomes the mother of all “internets”: the Internet is the big one, the one that everyone immediately thinks of when they hear the word. “Internet” is a proper noun, and therefore should be capitalized by the rules of the English language.

Admittedly, in common (non-technical) usage, it doesn’t make much sense to make that distinction. But this is Wired magazine we’re talking about, not the latest Hollywood gossip. It may not exactly be an assembly language reference manual, but it’s at least technically oriented, and should be managed and written with that in mind.

Now for my own personal doubts. Should “Internet” really be capitalized, especially considering how it’s used? I understand the reasoning behind it, I see why it’s been done that way so far, but is it really still necessary? Thinking back, I can’t recall actually using the word in its more generic sense; whenever I say “internet,” I’m talking about the Internet. And maybe that’s the important distinction: we can tell which internet we’re talking about based on context. “An internet” is a couple of networks hooked together, and “the internet” is the Internet.

As usual, I see both sides of the issue, and I’m not sure which side I favor at the moment. I think that, for now, I’ll keep capitalizing it out of habit, since that’s what I know for sure to be correct. Down the line, though, I might just follow the example of Wired News and stop capitalizing “Internet”…

Where Are My Bells and Whistles?

Just a quick gripe: C++ does not support constructor chaining. Grrr.

It isn’t a huge deal, and the page I found does offer a workaround which minimizes code duplication, but still… I think I shouldn’t have to write my initialization code in member functions which my constructors then call. Maybe that’s just me being overly picky. Maybe I’m just too accustomed to more “modern” programming languages.

For the curious and/or lazy, the workaround is this (although I kind of gave it away a minute ago already): write member functions in your class that do all the work of initializing a new object, and then call those functions in whatever combination a particular constructor requires. Not particularly pretty or elegant, but it works. The link I gave above gives a pretty detailed example, so if you want to look at exactly how it’s done, I recommend checking out the second half of that tutorial.

Ah well, back to my final project for Computer Science II. It’s going to be pretty cool, actually. Maybe I’ll post some screenshots and stuff once the semester is over and I’m not quite so busy.