Suffice to say, it is a Quad Core i7, 64-bit, 17″ screen monster.

My particular one has Intel’s 1.73ghz Core i7 Q 820 processor, which with hyperthreading gives you 8 cores to play with. 7200rpm drive for the main drive. ATI FirePro M7740 for the display adapter. 8G of RAM.

It’s, ah, fast. In a word.

Especially for something I carry around all over the place. Work, Barnes&Noble, home, kids’ dance studio, etc. A day does not go by without travel of some sort.

Wow, that’s big!
Make no mistake, it is a physically imposing laptop. It’s a big, vaguely ominous grey rectangular slab of computer. To wit:

• 8.4lbs/3.81kg
• Width: 15.4″/393mm
• Depth: 11.0″/280.5mm
• Height: 1.35″/34.5mm(F), 1.5″/38.5mm(B)

Carrying such a thing is interesting, to say the least. After it arrived I had to buy another backpack in order to transport it. (Sometimes I think I am running a shelter for abandoned laptop cases.) And it is heavy. That 8.5lbs feels like a lot more like 28.5 after a half hour. And the “Slim” power supply is still the biggest one I’ve ever seen, except for the non-Slim one I use at my desk: that one is actually the size of a brick. Yowza.

So the overriding impression of it just sitting around is easy to some up: Big.

Why so Much Laptop?
I could be cheeky here, and say you always need the fastest laptop, the biggest laptop, etc., but the reality is for a lot of people this is way too much power. This is a work-provided laptop to do work-related things on, and that means development and testing of a highly concurrent logic database product. I have no trouble at all taxing all eight cores and all eight gig of ram. To be fair, I have no trouble taxing the sixteen cores and 32 gig of ram on the servers at work either, so, it is what it is. But if the extra hardware gets me through compilation and testing 10 minutes faster, it is worth it.

Like I said, you might not need the power, but if you do, the M6500 delivers.

Why a 17″ inch Screen? Heat.
This is something I ruminated over for a while. In the end I didn’t actually want the 17″ screen, and would have been happy with the same 15″ screen my last two laptops had. But, what I really wanted was a physically larger form factor to cool that quad-core processor.

Having bought and used a lot of laptops at work, it will come as no surprise to most of you when I say heat is the biggest cause of laptop failures. Packing that processor in the 15″ form factor made me nervous. The other developer who was buying at the same time had the same thought but went with the Alienware M17x. It’s a nice laptop, and as fast as the M6500, but I think I did better. More on that later.

So the reason it is big is not that it had to be, but that I was conservative with the size in respect to the CPU power. My CPUs are running flat out a lot, and heat is a real concern. Just because a manufacturer stuffs a quad core in there doesn’t mean they want or expect you to use it flat out for hours at a time.

In that respect, the M6500 is superbly engineered. Yes, the fans run a lot when I stress the system, but the CPU temperature is well-controlled.

So it is All Good, Right?
Well, I love the machine. I love the speed, the bedrock reliability, the extra bay which currently holds an SSD (and which the Alienware does not have!), the more than adequate graphics performance. The dock Just Works, the machine has tons of ports, and it has my beloved eraser-head mouse pointer. (Touchpads suck).

All perfect, right? Just what you need in a luggable server class machine, right?

Well…

Keyboard Design of Doom
Really, this is too bad. And, apparently common, as I look at Asus, HP and other 17″ laptops at BestBuy or Staples.

Some twit decided that because there was the room, they should shove a numeric keypad on the laptop. We can debate the merits of a numeric keypad if you’d like, but really, I’d rather not. I use it some, but wouldn’t miss it.

But now I hate numeric keypads all over the world.

Because the damn keyboard is not centered on the machine!

And because the keyboard is off center, the mouse pad/eraser head is off center too. This on a machine that people use on their laps. A machine that weighs nearly 9 pounds. In order to use it you have two choices. Either learn to type at an angle, or have the laptop constantly slipping off your lap because you are using it slid to one side. And if the M6500 slips and falls, it will probably dent your floor.

A pox on the design team’s houses for this decision. It really screws up the ergonomics of using it as a laptop. And it never stops irritating.

Look around people. Apple doesn’t do this. Their keyboards arrive sans numeric keypad, squarely in the center of the machine. Like God intended.

Sheesh.

Other Nits
Other than the really amazing size of this thing, the only other nit I have is that the front lip of the laptop when open is a bit sharp, so much so that I had to invest in a wrist rest to avoid the sharp edge. Depending on how you use the the laptop, your chair and desk situation, this may be irrelevant for you. I’ve become very fond of the wrist rest in general though, and bought another to carry in my backpack.

Oh, and the battery life, as you’d expect, sucks rocks. But that is the deal when you take a quad core processor and beat the hay out of it — you burn through the battery in under two hours. No surprises here, about what I expected. I bought a mobile server, but i didn’t expect to run it off a battery.

Wrapping Up
So, in the main, I really like this laptop. I love how hard I can thrash it and have it stay cool, how it stands up to server use for hours at a time. The display is great, graphics are pretty good, and the durability has been great. The extra drive bay is a real plus.

The keyboard is a tool of Satan. Really, it is a shame, because the feel of the keyboard is great, but the off-center placement is an abomination.

Would I buy it again? Tougher call. I’d really miss the extra drive bay, and I’ll certainly never go back to mere dual cores, but I might chance the heat for a 15″ display and a keyboard in the center. I note that the M4500 has a quad core option with a second drive, but with lesser graphics. If I was buying today, it would tempt me. But I would worry about heat, oh yes I would.

This is given Apple a huge advantage, one that Apple’s competitors have failed to do anything about. And that advantage has nothing to do with songs, videos, apps or podcasts.

No, the real advantage of iTunes is that it ensures every iDevice user, first generation Nano to Verizon iPhone 4 to iPad, is an Apple customer.

I’ve owned an iPhone 3G and 3GS, both on AT&T (I moved onto a Droid X on Verizon). My daughter has a second generation iPod Touch (She moved on to an HTC Incredible).

It is hard to explain how much I loath iTunes-the-application. Slow, ugly, missing key config options (you try allowing gapless analysis on a 80gig music store residing on a network share on a slow network. Can’t turn it off, though). The idea that podcasts get downloaded to your computer, and then later synced to your phone? Prehistoric. The notion that I would connect my phone to my computer more frequently than once a month is nuts.

Some of my gripes are no doubt exacerbated by running iTunes on Windows. Apple may have $50 billion in cash reserves, but they’ve apparently never used it to hire some savvy Windows programmers, to judge by the pauses, crashes, and runaway CPU usage that iTunes on Windows is known for.

But dammit, having customers plug the phone (tablet, whatever) into iTunes at least every couple of weeks has an almost transcendent effect on the customer experience: you are reminded that you are Apple’s customer. When I had an iPhone I was an AT&T customer but I was only barely aware of AT&T, crappy call quality notwithstanding. Apple, Apple I was aware of. They cared about me.

Now, they might not care much. Certainly they do not care about me and my network drive full of personally ripped MP3′s, as I fall outside their common usage scenario. But still, when I plug the phone in, it syncs, and I get notified of new stuff. Like new firmware. New apps. Sometimes just messages. If I have a problem, I can plug it in and restore. It backs stuff up (takes hours, but it works). If I have a huge problem, iTunes is a reminder that there is a store nearby that will help me. Apple is out there, and they are more than vaguely interested in my user experience.

Contrast this with Android. In the Android world, there is no “phone-home” app. Now, avoiding iTunes for podcasts/audio/video is wonderful, but it bothers me that I am no one’s customer.

Who should be my overlord?

Let’s meet the candidates:

Verizon: We Are The Network
My phone is on Verizon’s network; but they don’t really care about me. And really, I don’t pay them for a user experience. I pay them for a pipe. For a phone call experience. There is no reason for them to care about how much I like my phone, they need to care about how much I like my phone calls. And they are great at this. So, no, I don’t really want to be Verizon’s customer any more than I am. Verizon gets a pass.

Google: Accidental Customers R Us
What about Google? Google authors Android, after all, and Google has the single largest impact on my day-to-day phone experience. It would make sense for Google to want me to be their customer. Right?

Well, Google doesn’t really ever have paying customers in the main. They make their money from advertising. That’s an experience where the customer doesn’t really participate in a seller/buyer style transaction. Google sells services to not-consumer customers. Have you ever tried to get someone from Google on the phone regarding Gmail? Good luck with that. Email functions as a “Great Wall of Google” when it comes to dealing with Google.

Further, in order to get Android on handsets on mobile networks, Google had to cede some of this ground to the handset manufacturers and network providers. I assume that is why handset owners like me don’t get updates from Google. Such a deal was probably worth it to get Android off the ground (and we could talk about Google’s attempt to market the Nexus directly) but…

Where Apple wins customers with smart design and glorious customer service, Google seems to want to attract users by dint of pure technical prowess. So for the foreseeable future, unless you are willing to by a Nexus series phone (on an inferior network), you are only accidentally a Google customer.

Motorola, HTC, Samsung et al: A Failure to Imagine?
Recall, I own a Motorola Droid X. My daughter an HTC Incredible. Neither of us gets our firmware updates from the manufacturer. I get mine from Verizon (someday). My daughter’s phone is not under contract, and I suspect I will need to root it to upgrade her firmware eventually.

It is not actually clear to me that either Motorola or HTC is aware we own these phones.

Why is Motorola so uninterested in having me as a customer? Why do they allow Verizon to sit between us like an uninterested crossing guard? Does Motorola have something personal against me? Maybe they dislike me for my Yankee fandom? Stranger things have happened.

By staying out of the customer relationship, Motorola ensures I have no particular reason to choose a Motorola phone next time. If the specs are better, I’ll buy HTC, or Samsung, or whatever. Almost four years after Apple rewrote the handset/network provider rules, all the other handset manufacturers are still happy to let Verizon/Sprint/AT&T/T-Mobile commoditize their products. Bizarre.

This prevents them from charging a premium for good products and prevents them from engendering any consumer loyalty. I bought the Droid X because it had the specs I wanted. I’ve had it for nine months now, and I have never had contact with Motorola. Even my warranty exchange was handled by Verizon. Why?

Don’t businesses generally want customers?

And pay Apple 30% for the privilege.

Peachy.

The early obvious candidate for this include the Kindle app, so popular in the iOS world. Amazon sells you a copy of a book for $10; they send (I believe) 70% of that to the author/publisher. $7. That means Amazon makes $3, out of which they fund their Kindle development, Whispernet maintenance, etc.

If they pay Apple 30%, well, they don’t have much of a business model, do they? The Nook and other eReaders fall into the same situation.

Apple bleats:

Our philosophy is simple—when Apple brings a new subscriber to the app, Apple earns a 30 percent share; when the publisher brings an existing or new subscriber to the app, the publisher keeps 100 percent and Apple earns nothing

From Ars:

… Apple’s terms for publishers offering newspaper, magazine, music, or video subscriptions is very clear: if Apple helps bring in customers, then Apple gets to take a cut of content sales.

See the truth being mongered there? No one would every consume any content on an iPhone if not for the iPhone. Apple bears 100% of the credit for any content accessed on an iPad.

Wow, that’s some towering hubris. Air thin up there?

Logical Conclusions

Ok, let’s roll with this a bit. E-Book Readers are used to access content bought elsewhere; Apple should get a cut. Electronic Magazines, another form of digital content, Apple wants, nay deserves a recurring cut.

Netflix is predicated on a subscription model. Does Apple really deserve 30% of my $8 a month? For delivering 500k of binary code to my phone, one time? Really?

Rhapsody/Pandora/Slacker et al, Apple just unilaterally changed your business model; did it feel good?

CVS has an iPhone app that lets you refill prescriptions; is Apple going to dip their fingers into that transaction? My healthcare?

You know, I buy my MP3′s from Amazon and load them onto my old iPhone via iTunes — does Apple want a cut of those purchases?

Wait, wait, wait! I buy shoes from Amazon via the Amazon app — does Apple want a cut of my shoe purchases?

I mean clearly I would never listen to an MP3, buy shoes, stream bytes, or read a digital book if not for my iPhone, right? Good thing Apple is out there, enabling me

And don’t think there is a substantive difference between digital content purchases and meat-space purchases. Buying a Kindle book from a link inside the Amazon Kindle app and buying a Kindle from inside Amazon’s shopping app both entail exactly the same overhead to Apple: None. If Apple can make a case that they deserve part of the sale proceeds from the former they can make that case regarding the latter.

Understand again, for those who might not be so technical, it costs Apple nothing when Netflix streams a movie to an iPad. Nothing. No resource cost, not overhead cost, nothing. It costs Apple nothing when you download a new Kindle book to your iPad.

In spite of this wave of Nothingness, Apple wants to be paid for it. Astonishing.

Out family spent something more than $2000 on Amazon last year; does Apple really want $600 of that, if I bought it all through the Amazon App? Really?

This kind of decision-making is so ridiculous that my 11-year old daughter lobbied for an Android phone (sans phone contract) to replace her iPod Touch.

Logical Conclusions the Second

So Apple has just made the App Store vastly less useful to large publishers, publishers with there own storefront and delivery system outside of AppleWorld. This is somewhat expected as nothing Apple has ever done has been for the sole reason to make their platform more useful for developers or publishers, and Apple would just as soon there is nothing outside of AppleWorld. Apple says they are in pursuit of the customer experience, but like any capitalist organization, that is a means to a financial end as well. As are most companies, they are in it to make money.

What’s it mean in the end for users and publishers? HTML5 gets to make a big comeback.

How long, really, will it take Amazon to make a fully fledged HTML Kindle reader? Netflix an HTML streaming app? Rhapsody? Pandora?

Longer than they’d like, but not forever. Some standards to push, some code to write, some pragmatic decisions to make. And if a bunch of them band together, spend some effort in building a common UI framework and such, all in HTML5, they could turn out HTML-based apps that come very close to matching native apps. Apple itself has done this in some cases, but the ease of the App Store pushed HTML mobile development down the food chain for the last three years.

Look for it to rise again.

What’s Apple going to do then, micro-charge for Safari Javascript execution time?

After ten years at the helm, Eric Schmidt is stepping aside, with co-founder Larry Page moving into the CEO seat. This will prove and interesting time for Google, no doubt.

Also interesting is Google’s current intellectual property death-match with Oracle over Android/Java issues. Feel free to read about this elsewhere.

This leads naturally to what Google really needs to do next.

Buy Java.

Seriously.

Google has somewhere north of $30 billion dollars cash on hand; Oracle paid around $7 billion for all of Sun.

Oracle does not really seem to interested in Java in and of itself; it’s just another technology that they can make money from. A further concern for Oracle is that Java is current fairly interwoven into their database products; in this sense buying Sun was somewhat of a defensive move.

Why Should Google Buy?
This is actually not just as simple as “Android”. Google was, until the Oracle gang rode into town, a big player in the Java ecosystem. Java is, to some greater or lesser degree, threaded all through Google, from GAE to GWT and so on. As long as Sun was a benevolent dictator Google had large tracts of the Java noosphere to play in. With litigious Oracle around, those lands are much less inviting.

Oh, Oracle would make Google pay. Steeply. But the goodwill Google would realize from such a move would be enormous. Google would be free to re-awaken Apache’s Project Harmony, for example. The Java ecosystem, kinda grinding along at the moment, would positively explode. As a side effect, the Android ecosystem would get another long-term injection of energy.

And Google would be able to cheaply push it onto every platform imaginable, with Android/Chrome/Whatever following. As in Android, Google isn’t after making money on the software, but in accessing the user experience. If Google puts Java everywhere, then it puts Google everywhere.

(Google would also no doubt be a much nicer refugee for loads of former Sun/Java engineers.)

Why Should Oracle Sell?
Well, first off, they could make a potload of money. Always a good thing.

More importantly, they are on a path to drown Java in litigation, arrogance, and uncertainty. I’m sure they are counting on someday reaping a per-handset fee from every Android handset sold, and that’s potentially a lot of money. I’m just as sure that if push comes to shove, Google will clean-sheet an entirely different language and runtime environment and toss Java out the door. There is no future where Google accepts paying a significant royalty to Oracle forever — once they have to pay royalties, they will leave it behind. And they will likely take the Java community with them, severely devaluing Oracle’s investment.

Assuming such a sale, Google would be happy to give Oracle all sorts of cross-licensing agreements regarding their own Java implementations and such. That would give Oracle peace of mind regarding their Java future.

And Oracle too would realize a lot of developer goodwill from such a transaction; the sighs of relief from Java developers around the world would be audible for days.

So…
Put it this way. Oracle cannot lead developers toward a particular technology. Google can’t help but do it. If Oracle makes Java hard, Google will just create something new, and Java will, eventually, wither.

Will Google buy Java? I doubt it. It makes too much sense to ever happen.

But Mr. Page, sir, if you’re listening, here’s an idea: Call Larry Ellison and buy Java it’s freedom.

I have, as has been noted before, a Motorola Droid X. Mostly, I use Grooveshark, Slacker, and Pandora for my music needs, but I do have 5000+ songs on a server at home. On my desktop (wherever I might be) I access those songs via Ampache, streaming as I need them.

To sync some songs with my Droid, I turned to the wonderful Media Monkey, under Windows. The Monkey is a truly awesome piece of software; the kids and I have used it forever to sync music into various MP3 players. Always works. Edits tags with ease. Displays none of iTunes despicable bloat. Doesn’t crash. Handles a network music collection with aplomb. My oldest daughter goes to sleep to a playlist in Media Monkey!

I’ve used MM for so long, on so many different little players, of course it was where I turned first for my Droid.

Trivial. Plug in with Mass Storage mode, configure the directories for Music and Playlists, boom, works.

I promptly forgot about it. I don’t sync that often. Local content is music of last resort for me.

Until this weekend, when I was recovering from my Number One Daughter’s ten-girl sleepover. Traumatic events prompt the strangest questions.

In this case I asked, “Surely I can sync my playlists and music from Linux, right?”

What a silly answer…

I went through three major players and a fourth, lesser contender.

Banshee
On track to be the default music player in Ubuntu, Banshee swallowed the networked collection (presented as a SSHFS mount point) with no problem. Created playlists. Saw the Droid!

Would only sync one playlist. Oh Bog, why do you torment me?

Really. One playlist. WTF? Little did I know that that was as close as I would get….

RhythmBox
The current default player in Ubuntu, RythmBox managed to grab my networked collection okay, though it gave the impression it was importing the music files somewhere else, which was scary. Plus, around 2500 files in, it crashed completely.

To get it to recognize my Droid as a mass storage device, I had to create a file in the root directory of my SD Card. Really. Magic! (The file in question is “.is_audio_player”). This does nothing, really, except make the Droid a place you can fling MP3 files willy-nilly. Not sync, just in-application copying.

That’s a fail.

Amarok
Running Amarok under Gnome is sketchy. You always feel like things are going to fly apart at any second. Plus it is part of the whole KDE4 wave, which drove me away from KDE in the end.

Amarok was happy to swallow the music collection. Fast too.

But, again, no real syncing. I couldn’t even get it to really give me a list of devices to configure. I suspect that it needs more of KDE’s framework to work correctly, but, you know, that’s not my problem.

And wow, what a contra-iTunes interface. Designed with the principal of Maximal Surprise, a trusted maxim of user interface design. Not sure I could live with it anyhow.

Fail.

Exaile
As a flyer, I took a shot at Exaile; which is small, fast, liked the music collection, but never saw the Droid, despite having a Device Configuration screen that made me hope. Sad face.

And so…
I’ve rebooted to Windows, rescanned my library (mounted via ExpanDrive/ssh), and re-synced. In less time than it took to tell you about it.

Perhaps I missed something obvious (always a possibility with Linux). I don’t really have time to futz with it anymore. I choose to view it as an experiment that didn’t pan out — that happens. But it is such a solved problem under Windows (Media Monkey is far from the only player that handles mass storage devices easily), and seems so straightforward, that you wonder just what is the problem? Why so hard? (Common Linux Refrain #7. On the other hand, I remember when auto-mounting mass storage devices was hard, so I have hope. Just not time.)

My only regret is that I didn’t try the developer build of Songbird for Linux. I just plum ran out of time.

The State of Linux Media Player Sync Support: Total Suck.

However, a couple years ago H2 changes it’s on-disk storage system. The new storage system is neater and cleaner, but it had a side effect: H2 is now significant less concurrent. Our deductive engine is highly concurrent (we implemented a form of Actor concurrency before we knew what Actors were), and we routinely run hundreds of SQL queries simultaneously. We treat the RDBMS backend much like a NoSQL database, and really thrash it at times, especially before the caching and materialization layers kick in. In recent years, as every machine gained cores like mad, this has become a problem. H2′s lack of concurrent execution sometimes slows the system down to the point where even on small datasets, Postgres is faster, despite the network hop. (This is not a knock on H2; highly concurrent access is not a design goal that seems to be emphasized. Fair enough.)

So this is a problem, so I decided it was time to survey embedded DB’s again. In the past, I had looked at HSQLDB and Derby and rejected them for performance reasons. Perhaps, in the intervening three years something had changed?

Derby: Sorry, Just Slow
Derby was pretty much like I left it: just too slow. Faster now, but still slow. It was very helpful to have SQL SEQUENCE’s; we need to generate a lot of unique identifiers, and sequences are perfect for this.

It took about four hours to get our 1700+ test framework to run over Derby. It worked fine. But it was slooooow. It was more concurrent than H2 (judging from the CPU usage), but it was just slower at getting to answers. I wonder if the whole SQL -> bytecode path is just flawed. I don’t know, and I am not going to dig too deep. Plus, Derby databases are slow to deploy, comparatively, and we create and delete about a hundred databases during a single test run, so that matters.

Finally, what a pain to programmatically manipulate. Shutdown by JDBC url? Set system vars to specify db location? All this stuff makes it harder to create, connect to, shutdown, and destroy multiple databases while in the same JVM. You can do it, but it requires a lot of careful management.

So while Derby was faster than last time I looked at it, it was still too slow.

HSQLDB
My other easy option was HSQLDB, which is all Java, shares some distant heritage with H2, and is quite popular. Plus, HSQLDB claims the engine is fully multi-threaded. Cool! Silver Bullet time!

Because HSQLDB, like H2, uses a fairly modern variant of SQL it was simple to deploy our system over it. In a couple hours I had the system running, and running quickly. Magic fast. It bothered me how fast it was.

Then I uncovered HSQLDB’s odd design choice of MEMORY tables. By default, HSQLDB keeps tables in memory and persists them to disk seperately. This leads to really fast execution (the data is in memory after all). But … our application needs the memory. Sad, but true. With H2, we our servers explicitly manage the amount of memory available for H2′s cache. Changing the default table type to CACHED moves the storage to disk. So I did that, and I likewise explicitly managed the cache size for HSQLDB.

Now the raw SQL performance was close to H2, just a shade slower in some cases, a shade faster in others . But if it was really concurrent, we’d get a big boost in deductive query evaluation. And indeed we did! For a certain class of query plan, the engine really flew.

Until everything hung inside HSQLDB.

Grrrr. Argh. Gnashing of teeth.

The hang was repeatable on my quad-core laptop under Linux (eight CPUs seen), but I couldn’t repeat it with a simple test case. The confluence of events and timing was just too complex to extract.

So I got the source, and tried it again, looking hard at the stack traces. Eventually I came to the HSQLDB class CountUpDownLatch, which implements a, well, counting up or down latch. This class has been around in HSQLDB a long time, and can be found in various source code repositories around the internet. The code looks like this:

import java.util.concurrent.CountDownLatch;

public class CountUpDownLatch {

    CountDownLatch latch;
    int            count;

    public CountUpDownLatch() {
        latch      = new CountDownLatch(1);
        this.count = count;
    }

    public void await() throws InterruptedException {

        if (count == 0) {
            return;
        }

        latch.await();
    }

    public void countDown() {

        count--;

        if (count == 0) {
            latch.countDown();
        }
    }

    public long getCount() {
        return count;
    }

    public void countUp() {

        if (latch.getCount() == 0) {
            latch = new CountDownLatch(1);
        }

        count++;
    }

    public void setCount(int count) {

        if (count == 0) {
            if (latch.getCount() != 0) {
                latch.countDown();
            }
        } else if (latch.getCount() == 0) {
            latch = new CountDownLatch(1);
        }

        this.count = count;
    }
}

What I saw, repeatably, was a hang in await(). As I looked at it I noticed two things.

First, for all that it is in a .concurrent package, it isn’t actually thread-safe. Most of the methods would fail miserably if they weren’t executed atomically. I ran around the source code for a while, but it looked like there was just one instance of the class, and that instance was always accessed via an exclusive lock. So, for all that CountUpDownLatch was not thread-safe, it seemed to be used in a thread-safe manner. As an aside, this happens via a ReentrantLock, not by a synchronized block. So, that is OK, I guess, though it goes against the grain that the code isn’t inherently thread-safe.

The second thing I noticed was subtler.

That ‘count’ variable — it kinda, sorta ought to be volatile. I would say it is the prototypical volatile use case. It is a variable whose value can be updated by another thread. The spec on volatile is here; a good explanation is here. Somewhat simplified, volatile is used to denote a variable whose value can be changed by another thread. If you don’t specify a variable as volatile, the compiler can cache the value (I believe in a register) and not go back to RAM every time. There are other events which cause this cached value to be flushed, but accessing a variable marked volatile forces the compiler to go back to RAM. (It is, of course, a good bit more complicated than this but the preceding is a useful working definition.)

In the code above, for example, if the value of count in the current thread was cached as 3, then another thread increments it to 4 , then the current thread decrements using 3 as it’s cached value, as 3-1 to 2, things are screwed up. Basically, the value of count is not guaranteed to be manipulated consistently, even though it is within the ReentrantLock.

So I wrote a bug, #3153989. But, you know, the code was just lying there in an Eclipse workspace, so I changed the variable to volatile, generated a jarfile, and rerun my tests.

Bingo, it all worked. Nifty.

And unsettling.

You see, the optimization of caching variable values into registers per thread is not one that has been commonly implemented; I believe newer JVM’s do it a lot more nowadays, particularly on server JVM’s on multi-core machines. That means that this code appeared to work fine on many machines, under many load scenarios. Only on a multi-core machine under concurrent load, with a JVM which does this caching optimization, could this show up.

— UPDATED —
Several posters cleared up some things. It seems that Lock.lock() does in fact flush the cache, thereby avoiding the problem in the manner I noted. Thus, volatile would not be needed, if every access to every method in CountUpDownLatch was under a lock. However, it turns out that the await() method in particular is often called outside of any locks, which is what subjects it to the volatile/caching issue.

So CountUpDownLatch is really specialized, and should probably be called “HSQLDBSpecificNotInherentlyConcurrentAtAllCountUpDownLatch”. Or some such.
— UPDATED —
The response to my bug was pretty swift, although the code available via SVN has not been updated yet. I suspect by the release of RC4, it will be.

But it did unsettle me — what other miscues like this are in there? So far I haven’t found any other problems, but we’ve got thousands and thousands of testing hours and end-user usage for H2. Obviously, it’ll be a while until I am prepared to trust HSQLDB to that level. That is not a knock on HSQLDB, just a fact of H2′s lead in our usage.

Wrapping Up
I did look at some other embedded RDBMS’s, notably SQLite, which is easily available from Java. But it is a much more primitive SQL, and thus lacks some things I’d like. And, well, it just wasn’t that fast with say, a hundred thousand rows. Not surprising, I am not sure that is the design goal for SQLite. Other, larger embedded DBs were not Java, and were going to be a royal pain to manage across 32/64-bit Linux/Windows/Mac.

So, we’ll probably at least put HSQLDB in the rotation for testing, and see how it holds up.

We also use, ABC.com, etc., to watch full episodes when our TiVo (rarely) gets snarled up.

I use MLB.tv to watch at least a hundred Yankee’s games a season; my girls are being raised properly.

We rent, early and often, from Amazon Video On Demand. I mean, it is our first stop for rentals; iTunes has been long banished, same as my first-gen Apple TV. Amazon Video On Demand just works: great quality, one-click purchase, easy searching, perfect.

I do all of this through a dedicated media PC, running Windows Vista with Home Media Center. I’ve got an ATI card with HDMI out connected to a fifty-inch DLP Samsung HDTV (which is getting old, but the picture is fine, so…). At the moment, there is nothing a dedicated media box like a Roku could do for me, so I am sticking with my PC.

I’ve got Boxee on the PC, and I use it occasionally. It is nice, but my kids prefer to just use Firefox and access all these services via the browser. Denigrate Flash all you want but the picture quality is superb. Even MLB.tv, which manages thrity-plus live streams fo HD material from all over the country to all over the country on a given day in the baseball season is pretty damn sharp. Interestingly, it got much better this past season when they dumped Silverlight for Flash.

And then, there is Netflix.

I pay the $9/month single-DVD rate; we tend to let the current disc linger for weeks at a time, but my kids love the Netflix streaming library. They use it on the Wii a lot, and on the media PC a lot more.

But … why does Netflix streaming quality suck?

I mean it is awful compared to every other service. Tearing is particularly prevalent, enough that it detracts from the experience. The overall quality at any given moment is fine, but the overall quality is lousy.

I note that Netflix still uses Silverlight for streaming. I make no further comment.

My internet connection is plenty fast enough, as I have no problem with anyone else’s streaming media. Heck, even 480p from YouTube is better — to say nothing of the videos available in 720p on YouTube, which are gorgeous.

For a company which terms itself “primarily a streaming content” company, their lack of investment in the streaming client is pretty surprising. You could argue that for several of the services I’ve mentioned, like the broadcast networks, or MLB, streaming is a sideline. Yet their quality is head and shoulders about Netflix. I know they stream a ton of video, I know they use a sophisticated dynamic-bandwidth methodology, etc. But…

The bottom line is, in my house, we rather stream from Hulu, Amazon et al, because the quality is just better.

That seems to be a problem to me. And for Netflix, I presume.

A scenario:

1. I have an email address.
2. You send me an email
3. I have, you know, a life. I don’t read your email for a couple days.
4. You get annoyed at me for not answering, so you send another email: “Did you get my email?”
5. I, still possessing a life, family and a job, still don’t read your email.
6. You get downright pissed, and send me a screed as to my personal failings.
7. Email checking finally bubbles to the top of my todo-list, and I read your first email. Possibly even respond to it.
8. By the time I read your last email, I am baffled as to how my personal scheduling became offensive to you.

Been there? Me too.

In order to understand how we got here, let’s take a quick trip through recent personal communication. I’ll skip the old caveman days when we grunted at each other…

Letters!
Long ago (forty years?), people wrote letters to each other. Regularly. On paper, written by hand. Writing a letter by hand takes some work; I encourage anyone to try it. Writing a three-or-four page letter to someone, in pen, to update them on your life, takes effort, time, quiet, and forethought. This is not a 140-character event. When you receive a real letter, well, you do feel an obligation to reply. At least in this case you know the person who sent the letter invested some time in it. It feels polite to reply.

The Phone.
Next up for personal communication was the land-line phone. Now you could call someone directly; dial a rotary dial, push some buttons and bang; for all intents and purposes an alarm went off in their house. Of course people felt compelled to answer; those old phones made noise! Plus, calling wasn’t a trivial thing to initiate, socially. You knew you were going to set off a loud noise in someone else’s house. This was a disruptive — almost innately rude — event. So people called only if they needed to. Which meant that if your phone rang it was probably important. So you answered it. It might be important!

Caller ID, answering machines, and simple familiarity started the slide away from this habit. Once we could see who was calling and be assured of not missing a call, well, then it became much easier to ignore the phone. Call during dinner? Forget it. Political call? Pass. And so on. In thirty years we went from phone calls being rare and important to needing a national Do Not Call registry. Progress, of a kind.

Sure, most people still answer the phone. Especially the older-than-thirty generation. But the trend was clear: you can ignore a phone call with little downside.

(As an aside, have you ever sat and ignored a phone call in front of some baby-boomers, or your grandparents? Drives them crazy. It especially seems to drive them crazy because it slowly dawns on them — do you ignore their calls too?)

Cell Phones
I’ll skip pagers, and move right into cell phones, particularly personal cell phones that you carry in your pocket or purse. Cell phones are wondrous and all, but they changed the nature of person-to-person communication pretty massively. First of all, it was the rare person who didn’t carry their cell with them; that meant they were always “home” to hear the ring. With tons of people carrying a cell all the time, the barriers to calling slowly dropped. Calling someone became as natural as turning to the person next to you and talking. Headsets, Bluetooth, voice-dialing, flate-rate long distance, mobile-to-mobile minutes; it all conspired to make it trivial to call and chat. For a lot of people, calling on the phone replaced listening to the radio when commuting. An incoming call on your cell no longer carried the weight of being even a little important, because we all became used to calling people just to chat.

But this ever-lowering barrier to calling people had a side-effect: it lowered the barrier to ignoring calls as well. Mix in a healthy dose of Caller ID for call screening, and people began ignoring calls, letting them go voice mail, figuring on returning the call later. And since you are always carrying your cell, sometimes it is just inappropriate to answer; kids acting up, in a quiet restaurant, in the, er, facilities. You understand. Society and custom conspired in several way to make us more likely to ignore your call. Maybe we could text you back later!

Texting was a real sea-change; texting was email to a phone. So let’s talk about email…

The Fickle Mistress, Email
Ah, email. We love our email. There was Compuserve, AOL, POP3, IMAP, Hotmail, GMail, Yahoo! Mail, Blackberry email, email on your iPhone, email from Nigeria, email about Viagra, and on and on. Email really changed the world; all of a sudden you could broadcast information to someone and they didn’t have to even lift a finger to get it! Magic. Soon, our inboxes were overflowing. Spam filters were born, and achieving “inbox-zero” became an Olympic sport for some.

Much has been written elsewhere about the broader social and societal impact of email, but I’m interested in one particular facet of email. Email added a new twist to personal communication: you sent me an email that is important to you, and you know it arrived at my inbox in seconds, but did I read it? You only know I read your email if I respond to it.

This creates a problem if you are a person who gets even a modicum of email. I get a fifty to seventy-five or so personal emails a day (those that make it passed GMail’s spam filters). I get several hundred work emails a day, mainly because our test servers generate email summaries constantly. Add in some email relating to the blog. Altogether it combines into a wave of email that I refuse to let dominate my life. This means I don’t always answer emails when I get them. Sometimes, with frequent emailers, I let several missives pile up and then answer them all with one email. So people who email me sometimes wonder if I read it. Which is, I’d say, their problem. Except that if the delay offends them, it becomes my problem.

(I’ll point out that a work email address most assuredly implies a social contract regarding answering your email. They pay you, you answer your email. In some cases, they hand you a Blackberry and expect you to answer it quickly. Fine, that’s the job.)

Email and texting are broadcast medium; fire-and-forget. Sure, with email you can ask for a read-receipt but nearly everyone will ignore it. Texting’s 140-character limit turns a broadcast service into a conversation service. The service itself isn’t suited to thoughtful exchanges of information, so people use it for short real-time exchanges or tiny informational updates.

Email, however, is well-suited to long form conversation. Quoting and threaded email make it easy to have long and involved discussions with people, spread out over time if necessary. Did you ever spend time on one of those mailing lists where they dissect a TV show or book series? Then you know what I mean.

Because I gave you my email address, did I make a promise to respond to your emails in a timely manner? Is an email address a social contract of some kind?

I think it is not, but loads of people think it is. And their idea of timely is their idea, rarely mine. So I offend people without even meaning too which annoys me. Now you can tell me I should just get to my email faster, but I have a life, a wife, kids, a job, etc. Time is my most precious commodity. I really do just prioritize other things ahead of my personal email.

Plus, I am pretty sure I am not alone in this. Because after email came…

Facebook and Twitter
Facebook and Twitter are (accidentally or not) designed to get around this problem by taking communication and making it asymmetrical. In Facebook you choose friends and post to your “wall”, which they can all then see — very efficient. Twitter is even looser. You follow some folks, but this relationship implies nothing about whether they follow you. Facebook’s “Friending” can be emotional for people (or at least “un-Friending” can be), but it is the rare tweeter that pays much attention to the comings and goings of followers.

I’d say the asymmetric nature of Facebook and Twitter communication is a big part of these services’ attraction. By removing a need for reciprocating, they let you talk without expecting a reply — with is both freeing and efficient. Sure, some people end up talking to themselves, but that is the nature of the beast.

But that returns me to my original question: is an email address some sort of promise?

I know mine isn’t. What about yours?

Note that these apps are not all free; after spending $200 for a phone and $60/month for the phone, spending $5 for an app doesn’t bother me. My logic, doesn’t have to be yours. The links are not links into the Marketplace though most should lead you there eventually. No order or ranking implied, and the prices are merely what I found on the net.

Launcher Pro
Free, fast, configurable, a good deal nicer than the default bi-directional home screen/launcher. I have it configured with seven screens, with the “home” one being the leftmost. Feels very iPhone-y when matched with a black background. Plays to how two years with the iPhone rewired my brain. Link

Wikidroid
Fast, free, intuitive interface to the Wikipedia. Since I use Wikipedia all the time for basic reference lookup, it’s a must. Link

Amazon & Amazon Kindle
I’m a big Amazon fan, and a reluctant Kindle fan. These are free (well, they both exist to facilitate the spending of your money, but still). Besides ordering, Amazon’s app provides a fast interface to Amazon’s reviews. Useful as all get out. Link and Link.

Evernote
Free, though I do pay the Premium fee because I use it so much. My primary on-phone note-taking platform. Cloud-based, syncs well, use it all the time. Link.

RealCalc
Free. I can’t use a regular calculator; I need an RPN calculator in the vein of my beloved HP-15C. (You can still buy an 12C for real!)
Link.

News (by smrtApps)
I don’t remember if this is free, but it is my favorite news aggregator. Link.

Handcent SMS
Free? I think? Either way, it is a step up from the standard SMS app. If you text a lot, you need it. Link.

K-9 Mail
Free. My day job email is IMAP, and I need to get to it fast, no fuss, with lots of support for folders. Link.

Hootsuite
$3. Worth it. Best Android Twitter client. Fast, actually does remember your position in the timeline. Pretty! Link.

Dolphin HD Browser
Ad-supported, or $5. Worth it either way if you like tabs with your browser. Plus, loads of plugins, fast, pretty, gestures if you like them, etc. My default browser. Link.

bTunes
$1.50. Awesome iPhone-like music player. Lock screen support. Headset control support. Works really, really well. Link.

TuneIn/Radiotime
One is free/ad-supported, one is $2. This is the best of the current crop of radio players, though Cherry RPlayer is as good a player with a lesser interface. Link and Link.

ESPN Radio
$5 I believe. Works damn well, pretty much identical to the iPhone version. If you are a sports radio junkie, it’s the ticket. Link.

Google Listen
Free. Podcasts are us. Interface is a little odd, but unlike BeyondPod hasn’t crashed for no reason. Link.

Pandora/Slacker
Free. I’ll lump these together, because they do similar things and I love them both. Pandora for the music, Slacker for the Adult comedy channel. There is something about standing in line at the grocery story hearing George Carlin’s Seven Words that just makes you laugh. Link and Link.

SoundHound
$5. Works as well as Shazaam, and you only pay once. Magic for the initiated! Link.

imov Messenger Corporate
$5? Whatever, it is the only actually working Jabber client that supports conference room chats. Period. Which I have to have. If you don’t need it, forget it, but if you do it is the only game in town. Yahoo and AOL both have IM apps for their respective networks, as does Google for Google Talk and they are all fine if you need them as well. Link.

DroidLight
Free, from Motorola! Useful, useful, useful. Link.

Tricorder (the one from moonbllink, NOT the execrable Tricorder-TR580)
This is super cool. In addition to the famed tricorder sound from Star Trek, it has a bevy of useful measurements including decibels, tilt, gps, compass, wireless and cell signal, etc. Link.

These are the ones I use regularly, and would immediately tell someone to install on a new phone, with the possible exception of the Jabber client, which is not for everyone.

Additionally, I have alarm clock apps, sport score apps, etc., but for most of these there are lots of good choices.

Droid-X Tip #1: Buy the OEM extended battery, and buy a hard case. Put the battery in, don’t put the phone back on, instead put the hard case on directly over the battery. Fits perfectly, stays slender. The extra 1mm of thickness of the battery perfectly matches the original battery’s thickness plus the back. Sweet.

Enjoy. And no, I still don’t miss my iPhone. Two months, and no dropped calls!

But do you remember what your parents (or perhaps your grandparents) reaction was? Do you remember how hard it was to teach them how to use the remotes? How you couldn’t easily explain that there was a separate tuner in the VCR, allowing them to record one channel while watching another? Some of you laughed at their confusion. Some of you no doubt cried in frustration.

In the end we are all tech support for someone, I suppose.

But do you recall the single defining visual of “old people” and their VCR’s?

I do:

Blinking Twelves! The Horror!

Today’s VCR (mine is 7 years old, but that counts as current for a VCR these days) sets the clock itself from a signal available on the cable feed. But for years, upon each and every power outage, the VCR would come back to 12:00, with the those digits blinking furiously, announcing your VCR was lost in time. And that you were unable to set it.

You could teach your parents and grandparents to use the VCR. I remember writing out step-wise instructions for my dad when I bought him an HDTV. Do you want to watch Hi Def? Follow these steps. Watch the VCR? Do this. Believe-you-me I made those instructions bulletproof, and it worked perfectly for him for years. I suspect his home theater setup in heaven has explicit instructions too.

But the clock, oh the clock! It would be fine for weeks, until the power failed. Few of us could remember how to set the clock after months and manuals were printed to be lost. No PDF’s in those days. And the clock was especially fraught in the olden days when you set the clock NOT via the remote but using an opaque sequence of tiny button presses on the face of the VCR. Sometimes the buttons were even hidden behind a little door. Lovely.

If you mostly used your VCR to watch rental movies (do you remember Blockbuster?), you didn’t even need the clock! I knew of people who taped over the display so the shame of a blinking clock would go away. Other folks just set the time to 12:01 and let the clock be wrong all the time. Madness in support of confusion!

Our parents and grandparents are not stupid; why did they not learn to set the clock and program the VCR themselves?

I have a theory.

Eventually, if you are not careful, it’s easy to get tired of learning.

At some point, “grown-ups” were just too dang busy to spend the time to learn how to do something. To learn such a thing requires adding yet another abstract model to the brain, another set of instructions and rules to be remembered. This is daunting for anyone who worked all day, or is a little older and tireder, who chased kids around all day, for anyone out there living their life. So don’t excoriate them for this.

Besides, you’ve got Blinking Twelves of your own.

Yes, you.

What are your Blinking Twelves, you ask?

Well. We live in a time where technology explodes at an amazing pace. All around us, new technologies flourish, become important, and die on the vine in what seems like flashes of time.

I started with computers in Junior High (what now seems to be Middle School), acquiring a succession of Timex/Sinclair, Radio Shack and finally Apple computers in the early 1980′s. Modem’s came along, bulletin boards were cool! Bought an IBM PC for college; doing my class projects in Turbo Pascal was much easier than using the mainframe. Jumped into C, released some shareware (still out there!), learned more Unix. Cell phones, Windows, OS/2, Linux, HDTV, TiVo, VOD, smart phones, CD, DVD, mp3 swapping, e-Books, Bit Torrent, and so on; all these things marched into (and in some cases out of) my life.

But you know, I just bagged MySpace and Facebook. I like LinkedIn because it is, well, less social, and takes very little effort to maintain. It’s actually useful career-wise. But Facebook and MySpace struck me as a lot of effort to bare my life to other folks. There’s no one whom I want to know that much about me, or who I want to know that much about. My wife, I see her everyday. Same for my kids. My colleagues. Phone calls to Mom, and my sister. Mostly, that keeps my life full.

So when MySpace, and then Facebook started exploding in popularity, I just sort of shrugged. Ho-hum.

(Interestingly, I love Twitter. Really, really love Twitter. It hits some deep sweet spot for me. I follow some politicos, some actors, and most importantly, a bunch of Yankees writers/fans/etc. Twitter is a great companion when watching a game. And I completely control the investment of time. If I feel like ignoring my timeline for a while, fine. No one cares. Someone tweets too much, unfollow them. Simple. Elegant. B’gosh do I love simple and elegant.)

Similarly ignored technologies abound. Mostly, I completely ignored the Java web application world most of my professional colleagues are enamored of (and earn a living using). Google Wave? Seemed like a lot of work. Google Voice? Looks nice, but … maybe later. Blu-Ray? Not yet, maybe someday. YouTube? Rarely. Flickr? Once, twice, that’s it. StumbleUpon? Delicious? Never. Chatroulette? Now you’re making me laugh. Hard.

For all these new capabilities, the point is simple; I consciously chose not to learn any of them; most of them I chose not to even investigate them and see what their use might be.

I chose not to learn to program those VCR’s.

Now, that decision is OK if I did it with my eyes open. But I never really played with Facebook, or MySpace, for example. And if you are deriding new technologies as worthless because they make you uncomfortable, or seem “too hard”, well, eventually, you’ll be left staring at the Blinking Twelves. And people will laugh.

The Internet, by the way, was the most hugest VCR of my lifetime as far as whole industries were concerned. Newspapers, movies, records, cable companies, radio stations, brick and mortar retail, and countless other industries had market leaders (if not entire markets) crushed because they were afraid of the Internet; they were afraid to learn to program that VCR.

Sad, or not? Got me.

Either way, the important lesson is that there is always a new VCR around the corner. Resist the impulse to throw up your hands and refuse to learn how to program it. Once you stop learning these newfangled things you might just fall off the progress bus. That would suck.

Me, I’ll be heading over to Facebook. Hopefully it’ll be interesting.

What are your Blinking Twelves?