Today I did some big mass emails via Amazon SES. Delfin released a new music video, and I had to send a press release to about 1650 journalists’ inboxes and to 6500 inboxes from the fan base. A few numbers and a slap on the wrist from Amazon below the cut.

I started the campaign from an account specifically set up for the task, registered to, let’s say, the management. The account was new, so it had a starter limit of 1000 emails per day and 1 email per second, which obviously wasn’t enough. After hitting that limit I switched to my own account, which is already up to 10,000 emails per day and 5 emails per second.

The list had been built up over a long time of work. The result — out of 1650 inboxes, 240 emails bounced. Almost all of them — addresses that no longer exist. About 10 came back as spam, many of them on the same domain. Several bounced because the recipient inboxes are full and refuse new mail, which in my view is essentially the same as being dead.

Next came the list of 6500 — registered users of Delfin’s fan site (I don’t think they’ll mind, the news of a new video is a pleasant one, and a mailing like this happens at most twice a year). Some of those, of course, are spam bots that stalled at the email-activation stage. I have no idea how many bounced from that list — at 410 the inbox on Masterhost (where all the bounces and error messages were collected) glitched out. It stopped accepting new messages, and refused to even open some of them.

The result — 30 minutes ago I got an email from Amazon that says:

Dear Amazon SES Customer,

Our analysis of your Simple Email Service sending pattern is showing that certain parameter(s) of your sending are at unacceptable levels. If these metrics do not improve, we will suspend your SES sending privileges. For each metric that is at an unacceptable level, we will allow you a defined number of emails to be sent to bring the metric back up to acceptable levels. Please note that if your sending quality decreases further actions may be taken before you have sent the number of emails noted below.

*Unacceptably High Bounce Rate: 1,414 of the last 11,000 emails you have sent have bounced.
*This issue must be fixed within the next 40,000 emails you send.

So they basically wagged a finger at me — 1,414 emails bounced and the sending quality is suffering. Don’t fix it, and we’ll throttle you.

Looked at the stats — 7,429 emails went out from my account in 24 hours. Which means about 19% of the messages were sent in vain and threaten sanctions.

The takeaway — if you’re doing (legitimate) mailings, you have to keep your address list fresh. It saves nerves and time later.

Installed the new Windows Developer Preview in VirtualBox. It started up on the third try, but it did. The new Metro interface gave the impression that it’s entirely aimed at tablets and finger control.

It’s wildly slow on the VM in places — though I only allocated it 1 GB of RAM.

The boot screen is modest:

After loading you see this beauty:

Drag this screen up with the mouse and log in on the green screen:

And here’s the Metro interface’s start screen. It’s made of widgets that surface information from their apps.

Tapping Desktop drops you into the familiar Windows 7 / earlier interface. The first thing you notice is a different Start button. Pressing it doesn’t open the usual menu — it takes you back to the Metro start screen.

The new Explorer with the ribbon interface is, to me, simply hideous. Microsoft argues that this interface is the result of some research, and that everything is more convenient and clearer for users this way. I have serious doubts. Although maybe I’m spoilt by the minimalism of Finder on Mac OS X.

The system ships with Internet Explorer 10. In the familiar Desktop mode it looks familiar. I didn’t feel particularly motivated to dig into it.

If you go back to the Metro interface and open Internet Explorer there — it looks more like a tablet-oriented app. And visually it fits the interface.

Back to the widgets. The first thing I launched was Tweet@rama. Logged in. For some reason the app didn’t load the timeline — only the list of my own tweets. Scrolling further right shows the list of followers and people I follow.

It’s only a list, though. Clicking a person’s username does nothing. Only clicking your own name in the list of your tweets opens up your account page on twitter.com inside the embedded Internet Explorer. Why like that — unclear. Clearly not finished.

Next — Socialite. Social-network integration. So far it only integrates with Facebook. Maybe nothing else will be added. It looks pretty effective. The home screen lets you pick News Feed, Profile, Photos, Friends and Checkins. The names speak for themselves.

News Feed:

Profile. On the right — the user’s recent activity:

An album in the Photos section:

The Friends section is, I think, very well organised:

Next — the Headlines widget. The icon makes it clear it’s an RSS reader. Each subscription is on its own screen (again, like everywhere here, with horizontal scrolling). It shows the latest entries by default, but you can load more. Newly loaded entries extend the current screen further to the right.

That interface is eerily reminiscent of the Pulse News app for iPad.

Another app — Stocks. Nothing extraordinary, all in the same pleasant style. Just charts and information.

Weather is done very nicely — an animated background depending on the weather; if it’s raining in your city, you get a rain video as the background. iPad-app vibes again. Temperature defaults to Fahrenheit, by the looks of it. I couldn’t be bothered to find where to change it.

The Control Panel is also done in tablet-app style:

There’s a set of standard sample apps too — like a jigsaw puzzle game. Nothing particularly interesting though.

The Near Me app, judging by the name, is supposed to be tied to geolocation and find people or places nearby. It refused to launch on the VM, crashed and dumped me back on the start screen. The same happened with a couple of other apps.

There’s a mopod app — you can subscribe to and listen to podcasts in it. There’s a built-in subscription list. You can also add your own podcast by URL.

There’s also picstream — uses the Flickr API and, as the name suggests, displays photos from there as a slideshow. By default it shows what’s popular on Flickr. You can also pick tags, but that didn’t seem to work for me.

I expected to see apps for listening to music or watching video — strangely, I didn’t find any.

Overall the interface is cool. Switching between the classic and Metro UIs feels rather jarring. On tablets I think it’ll look great, especially since Microsoft has already road-tested a lot of this on Windows Phone 7. The main thing now is that there be software.

The weekend was spoiled by my MacBook Pro: it had gone to sleep at the office, woke up at home, and a few seconds later froze solid. Had to do a hard reset.

The OS refused to boot — neither from the main partition nor from the Recovery HD partition that OS X Lion creates for restoring/reinstalling the system. Although the MacBook could see both partitions. I managed to boot from a USB stick.

My first thought was something wrong with the hard drive — either partitions had detached or the disk was dying. I pulled the drive, hooked it up to my second Mac via a USB enclosure. All files in place, no signs of trouble. OK, I figured, time to reinstall macOS from scratch — looked like a software glitch, those upgrade-from-previous-version updates can be flaky.

Backed up everything, ran Disk Utility’s repair, it fixed something I didn’t understand. Stuck the drive back in — decided to try booting from it — and lo and behold, everything worked. I breathed a sigh of relief, shut down, started up, slept the laptop, woke it — all fine.

It sat quietly until the next evening. I opened it — and the same thing happens — freezes and won’t boot any more. Booting in verbose mode (CMD+V at startup shows the boot log on screen instead of just the apple on a grey background) showed me some unfamiliar errors, clearly disk-related — the system couldn’t mount one of the partitions. This time it wouldn’t even boot from the USB stick. I pulled the drive again, wiped all partitions — no good. Stuck in a different drive — and that’s when it turned out that one was sometimes invisible too. Annoyed, I drove to the service centre, by then almost certain the problem was with the cable.

That’s exactly what it was — they swapped the cable, charged me 2,500 roubles. They ran my MacBook through their tests — the rest of the hardware is fine.

Investigating the cause of the cable damage is unlikely to be possible. But the cable looks very strange.

On top — the new one, freshly installed. Below — the old one, all marked up. The cable itself is quite thin, and those pinpoint dents could easily have broken contacts inside it. What’s strange — the same dents are on the optical drive, and in a few other places. And all of this is inside the case.

Another shot of the old cable

The cable runs through the case the way you can see in the photo above (the hard drive lives bottom-left), with the case lid sitting on top.

The only theory I have for what could have happened: sand got inside, kept shifting around in there and getting pressed when the laptop was being carried in a backpack. In principle it could have got in in Vietnam, where I spent a month and occasionally got the laptop out on the beach — and on that beach there’s usually a strong wind (kite-surfers come there for a reason), with sand constantly in the air. But that was a year and a half ago — why the problem only showed up now, and where the hell all the sand has gone, given that it was inside a closed case, is still a mystery.

Now I need to give the laptop’s insides a proper cleaning, in case anything’s still in there. And tape over that cable with something. They suggested heat-resistant tape, the kind that doesn’t mind temperature. Just to put a layer between the cable and the case lid, so that if anything gets in again, it scratches the layer rather than the cable. Maybe a thin foam strip would work too — we’ll see when I get to it. Sand is evil.

By the way, this cable plugs into the motherboard with this odd-looking thing:

Needed to mount a drive with the HFS+ file system (the one Mac OS X uses). It’s pretty easy. Ubuntu’s repos have a package called hfsplus.

sudo aptitude install hfsplus
sudo mount -t hfsplus /dev/sdf2 /home/username/folder

To unmount:

sudo umount -t hfsplus /dev/sdb2

Hooray — got myself a new body, the Canon 60D. I’d been wanting a 7D for a long time, but eventually, thanks to @theproof and a couple of other people, it turned out the 60D has the same internals as the 7D — same processor and same sensor. The 7D just has two processors, which only really affects burst-shooting speed and how many frames in a row it can manage. A few photos taken on the way home, below the cut.

The flip-out screen is, of course, a handy thing. You can also fold it inwards so it doesn’t get scratched in a bag. Phone snap:

Test shots (clickable):

Another post — to help things settle in my own head. About working with JSON in Objective-C, using parsing of tweets from Twitter’s public timeline as an example.

Out of the box Objective-C has nothing for working with data in JSON format. Only with XML. But there is a JSON framework, in case you’re forced into this by necessity / whim / your boss’s refusal to give you the data in XML (underline the relevant option). You can download it on GitHub. The latest version at the moment is 3.0.3. There’s also an alpha 3.1, but that’s an alpha.

The archive contains a Readme, examples, the framework itself, and some other files whose purpose is so far unknown to me :)

The framework files themselves live in the Classes folder, which is what you need to drag into the file list of your Xcode project. For convenience I renamed the folder to JSON after importing.

In RootViewController.h we import the framework with the line:

#import "SBJson.h"

In RootViewController.h we add the following code:

@interface RootViewController : UITableViewController  {
    NSMutableData *jsonData;
}
 
@property (nonatomic, retain) NSMutableData *jsonData;
 
@end

We declare a class property jsonData of type NSMutableData — that’s where we’ll stash the data returned by the request to the Twitter API.

The property line is needed in case we want to access the jsonData property from outside our class. nonatomic means that jsonData isn’t locked, i.e. we can both get and set the value. They write that nonatomic is faster than atomic. Atomic means we lock the value. About retain they say: How the setter method will set the variable. I haven’t fully figured out yet why this is needed and how to work with it, but without that line nothing works =).

We’ll grab JSON data from Twitter. Conveniently, the public timeline doesn’t require any auth or tokens — you just send a request to a particular URL, perfect for trying things out.

To start with — we have a class RootViewController. In RootViewController.m we’ll put the following code in viewDidLoad:

@synthesize jsonData;
 
- (void)viewDidLoad {
    [super viewDidLoad];
    // create an NSURL object with the address the request will go to
    NSURL *url = [ NSURL URLWithString: @"http://twitter.com/statuses/public_timeline.json" ];
 
    // create the NSURLRequest — the request itself
    NSURLRequest *theRequest=[NSURLRequest requestWithURL:url
                                              cachePolicy:NSURLRequestUseProtocolCachePolicy
                                          timeoutInterval:60.0];
    // start the connection
    NSURLConnection *theConnection=[[NSURLConnection alloc] initWithRequest:theRequest delegate:self];
    if (theConnection) {
        self.jsonData = [NSMutableData data];
    } else {
        NSLog(@"Connection failed");
    }
    [theConnection release];
}

To handle the connection and the data, you also need to add this code:

/**
 * As each new chunk of data arrives, append it to what we already have 
**/
- (void)connection:(NSURLConnection *)connection didReceiveData:(NSData *)data {
    [jsonData appendData:data];
}
 
/**
 * If the connection fails — log the error to the console
**/
- (void)connection:(NSURLConnection *)connection didFailWithError:(NSError *)error {
    NSLog(@"%@", error);
}
 
/**
 * Once all data has arrived — parse it 
**/
- (void)connectionDidFinishLoading:(NSURLConnection *)connection {
    // store the received data into the result string
    NSString *result = [[NSString alloc] initWithData:jsonData encoding:NSUTF8StringEncoding];
	
    // we can log it to the console and inspect what we got
    NSLog( @"%@",result );

    // create the JSON parser object
    SBJsonParser *jsonParser = [[SBJsonParser alloc] init];

    // parse the result string into an NSArray
    NSArray *dataObject = [jsonParser objectWithString:result error:nil];

    // breakpoint can go here

    // don’t forget to free the objects
    [jsonParser release];
    [result release];
}

If you set a Breakpoint (CMD+\) at the spot indicated above (see «breakpoint can go here»), you can inspect the values at that point. The screenshot below shows that the string we received from Twitter was parsed into a dataObject of type NSArray, which contains 20 items inside it — i.e. data on 20 tweets from the public timeline.

After the breakpoint line, the received data can already be processed somehow. For example, you can log the received tweets to the console as «Username — TweetText»:

for ( NSDictionary *tweet in dataObject ) {
    NSLog(@"%@ - %@", [[tweet objectForKey:@"user"] objectForKey:@"screen_name"], [tweet objectForKey:@"text"] );
}

Hopefully I haven’t lied or got anything wrong anywhere. Corrections / clarifications very welcome.

Why not pin this here — easier to find later, when I need it again. Search engines often treat pages like

arm1.ru/blog/yandex-upal-panika-v-twitter

and

arm1.ru/blog/yandex-upal-panika-v-twitter/

as different pages. So you end up with the same content technically living at two URLs, which isn’t great. Below the cut — code that automatically strips a trailing / via a 301 redirect, so search engines don’t end up with duplicate pages.

<?php

# strip the QUERY_STRING from REQUEST_URI
$uri = $_SERVER['REQUEST_URI'];
if ( false === empty( $_SERVER['QUERY_STRING'] ) )
    $uri = str_replace( '?' . $_SERVER['QUERY_STRING'], '', $uri );
			
# 301 redirect when there is a trailing slash on $uri
if ( substr( $uri, -1 ) == '/' && strlen( $uri ) > 1 ) {
    $queryString = '';
    if ( false === empty( $_SERVER['QUERY_STRING'] ) )
        $queryString = '?' . $_SERVER['QUERY_STRING'];
 
    header( 'Location: ' . substr( $uri, 0, -1 ) . $queryString, true, 301 );
    exit;
}