A couple of people in both forums and IRC have recommended the Acer Aspire Revo 3610 net-top machine as a candidate for running Boxee, so after a little research I picked on up from eBuyer at a bargain £180 for the N330 / 2GB / 250GB / Linpus configuration, delivered the next day for free. After firing it up for the first time, I spent less than 5 minutes poking around the pre-installed Linpus out of curiosity (the 250GB disk was oddly configured with a single 10GB partition), then promptly wiped it and installed Ubuntu 10.04 (x64) and the latest Boxee release.

Having previously read Liam Green-Hughes’ website, I was aware that I needed to tweak some settings to get the digital sound working, though there was a slight difference with my Revo in that I have three different S/PDIF levels within alsamixer – “S/PDIF”, “S/PDIF Default PCM” and “S/PDIF 1″. I un-muted all  of them but have yet to experiment to see what each of them does. I set my output device to “Digital Stereo (IEC958) Output” within the Ubuntu sound configuration as I want the audio to be sent to my amp over S/PDIF for decoding; after messing around with a few different combinations of settings within Boxee I was getting sound for files encoded with normal stereo sound, but silence for anything with multiple channels (DD / DTS). After a couple of hours of tinkering and research, it turns out that pulseaudio has a default sample rate of 44.1kHz and AC3 (Dolby Digital) is signed 16-bit little-endian 2-channel 48kHz PCM data, configuring pulseaudio to run with a sample rate of 48kHz appears to have fixed the issue, as I am now getting DD and DTS passed to my amp for every media file I have tested.

After playing around with the Revo / Boxee combination for about a day, it’s generally all positive; it plays back 720p/1080p content perfectly by passing the majority of the HD video decoding to the ION chip with the help of VDPAU available within GNU/Linux, the CPU is running at ~10% during all of this while managing perfectly with non-HD content too but with a little more strain on the N330. While “idling” on the Boxee home screen, one of the CPU’s threads is curiously always running at 100% load, though it is possible it is still indexing all of my media with it reporting “identifying” next to one of my sources. It has currently successfully identified 99% of my media requiring me to manually identify a handful of files, but this is perfectly acceptable considering the brilliant interface complete with covers and descriptions for all of my movies and TV shows. There is a slightly annoying wait when switching between the different areas of Boxee and sometimes a slight bit of lag when browsing through the media, but this could be down to it still indexing the content.

There appears to be plenty more to play with such as web content and applications, from what I have looked at, Flash video plays back fine in standard definition but HD Flash content is completely unwatchable – This is likely down to the lack of 64bit Linux support from Adobe. Other web based content such as the video made available within the Revision 3 application plays back perfectly, I’m pretty sure it’s HD too, but I couldn’t find an option to switch between the different encodings of their content.

Instead of mixing in posts about Forex on this geekish site, I have set-up a separate one – Starting Forex – that I’ll be using to document my findings and things that I have learnt, mainly to remind myself of stuff, and I find I get a better understanding of a subject if I write down what I learn.

I decided to use the built-in cURL functionality of PHP for interacting with the APIs. This code uses some of the JSON functions introduced in PHP 5.2.0 so a recent install of PHP is required as the Bit.ly API responds with a JSON string. On to the code…

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53

function shorten_url($long_url) {
    $login = "username";
    $api   = "api key";
    if(!empty($long_url)) {
        $shortener = "http://api.bit.ly/shorten?version=2.0.1&longUrl={$long_url}&login={$login}&apiKey={$api}";
 
        $ch = curl_init();
        curl_setopt($ch, CURLOPT_URL, $shortener);
        curl_setopt($ch, CURLOPT_HEADER, 0);
        curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
        curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
        curl_setopt($ch, CURLOPT_TIMEOUT, 10);
        $x = curl_exec($ch);
        curl_close($ch);
 
        $json = json_decode($x,true);
        return ($json['statusCode'] == "OK") ? $json['results'][$long_url]['shortUrl'] : false;
    } else {
        return false;
    }
}
 
function twitter($status,$url) {
    $username = "username";
    $password = "password";
    if(!empty($status)) {
        $limit = 140;
        if($short = shorten_url($url)) {
            $length = ($limit - (strlen($short) + 4));
            $status = (strlen($status) > $length) ? substr($status,0,$length)."... ".$short : $status." ".$short;
 
            $ch = curl_init();
            curl_setopt($ch, CURLOPT_URL, "http://twitter.com/statuses/update.xml");
            curl_setopt($ch, CURLOPT_HEADER, 0);
            curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
            curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
            curl_setopt($ch, CURLOPT_TIMEOUT, 10);
            curl_setopt($ch, CURLOPT_USERPWD, "{$username}:{$password}");
            curl_setopt($ch, CURLOPT_POSTFIELDS, "status=".$status);
            $x = curl_exec($ch);
            curl_close($ch);
 
            $p = xml_parser_create();
            xml_parse_into_struct($p, $x, $vals, $index);
            xml_parser_free($p);
            return ($vals[0]['tag'] == "STATUS") ? true : false;
        } else {
            return false;
        }
    } else {
        return false;
    }
}

There are a few small issues that don’t make this reconfiguration as straight forward as you would hope. First of all, the required script (mt-daapd-ssc.sh) is not installed along with mt-daapd, so I had to pull it from the svn repository and drop it into /usr/bin. Also, mt-daapd-ssc.sh uses wavstreamer to encode the decoded FLAC file, so if that isn’t available on your system, you’ll need to install that too. I believe it is built along with the rest of mt-daapd when building from source.

Next up, mt-daapd.conf needs to be modified to change how the transcoding is handled, the following lines need to be uncommented:

ssc_prog = /usr/bin/mt-daapd-ssc.sh
ssc_codectypes = ogg,flac,alac

The following line needs the ssc-ffmpeg.so part removed:

plugins = rsp.so,ssc-ffmpeg.so

At this stage, I thought that this was all that was required, but I still couldn’t get it working. Upon searching the Ubuntu Forums, someone else had already been through this and pointed out that one component of this set-up (I’m unsure which) requires an older version of FLAC, so a symlink had to be created.

sudo ln -s /usr/lib/libFLAC.so.8.2.0 /usr/lib/libFLAC.so.7

After restarting mt-daapd, iTunes was again playing FLAC files!

Our new printer, an HP 2727nf MFP, was a completely different experience. After hooking it up to the network and selecting to add a new printer within Ubuntu, it searched for printers, found the new HP, fetched and installed the required drivers and just worked! Pages are sent to the printer almost immediatly and have printed perfectly every time, so far.

Also, I set the same printer up on 6 machines in the office, 1 x Ubuntu, 2 x Xubuntu, 1 x OS X, 1 x Vista, 1 x XP Pro. By the time it was set-up on the first Windows machine I had already completed the set-up on all three Linux machines – Two thumbs up for Ubuntu and HP.

However, browsing the root of the Trac install does give a list of all the available environments, and there are RSS feeds available for any of the reports that have been created. So I wrote a quick script to pull information from both of these resources and display a list of all available environments and how many open tickets each one has.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
    <head>
        <title>Available Projects</title>
    </head>
    <body>
        <?php
        require_once("magpierss-0.72/rss_fetch.inc");
 
        $root = "http://trac.kris";
        $c = file_get_contents($root);
        preg_match_all("/<a href=\"\/([a-z0-9_-]+)\".*>(.*)<\/a>/i",$c,$matches);
 
        echo "<h1>Available Projects</h1><ul>";
        foreach($matches[1] as $k => $v) {
            $rss = fetch_rss("$root/$v/report/1?format=rss&USER=kris");
            echo "<li><a href='$root/$v/report/1'>".$matches[2][$k]."</a> (".count($rss->items).")</li>";
        }
        echo "</ul>";
        ?>
    </body>
</html>

After taking it apart and hooking it back up to some power, a guitar and an amp, in its disassembled state it was working fine – Reassemble and the fault reappears. It was instantly obvious that something was shorting out against the metal casing. You’d have thought that there was some form of non-conductive coating on the base-plate – Maybe there is, but defective near the switching area of the circuit. Covering the base-plate with insulation tape fixed the issue and it’s now working perfectly again!

Upon booting there was no sign of the cards BIOS during POST and no sign of the disks attached to it either after booting into Ubuntu, though running lspci reported that the card was present. After poking around with a mixture of Google and the Ubuntu forums, a number of people were having issues with the card but no conclusive fix had been found.

So I head on over to the Silicon Image website to check if there are any new BIOS images available, and surprisingly there was! After downloading, and realising that the only flash tool available was DOS based along with my server not having a floppy drive or a CD drive to boot into a DOS shell with I managed to get a USB stick booting into DOS with the flash tool.

To cut the story short – Flashing the card with the latest non-RAID version of the BIOS fixed my issue. It was now displaying the BIOS upon boot, and the attached disks were visible when running lshw -C disk.

After doing a little research, it turns out that the mic is surface mounted to the keyboard PCB so to replace the mic I’d need a new (or working) keyboard PCB. The only place I could find this component new was on some American website, which after shipping and other taxes involved in importing stuff, would have cost about £70. Or, I could have sent it off to Vodafone for repair at a cost of £60 and a few weeks Blackberry-less no doubt.

I ended up grabbing what was described as a faulty 8800 from someone on eBay for about £40, the description said it had a software fault, so there was a good chance that the hardware was in good condition. It turns out that there was nothing wrong with it from what I could gather after messing around with it for a couple of minutes. But it was on the o2 network, so no use to me.

After disassembling one of my 8800s it turns out that the keyboard PCB is somehow permanently fixed to the chassis. I made several attempts to split the two apart, but was concerned as to how functional the PCB would have been afterwards, and how I was going to mount the replacement. I’m assuming that if I were to order the replacement part from America, it would arrive mounted to the chassis. The only issue with this, is that the sticker with the serial number, IMEI, BT MAC etc. is also permanently stuck to the back of the chassis.

There isn’t too much work involved in taking one of these things apart. After removing the battery, MicroSD card and sim card out there are four Torx T5 (or T4 – My precision Torx drivers aren’t marked) screws to remove. After removing the two chrome plastic side plates, which are just clipped on, the back cover comes off with a bit of jiggling.

After removing the back cover, there are a further two T5 screw to remove near the top which releases the clipped on front cover and keypad. At the bottom there is then a plastic cover that just pulls off, underneath is a small PCB which looks like it contains some form of antenna, possibly GPRS/EDGE based on what was printed on the cover.

The next task is to remove the frame that the loudspeaker is mounted to. This frame is just clipped on with two clips either side, and two clips at the top. There is also a wire linking the main board to the antenna board which is clipped to the frame, it’s easiest to unplug the wire at each end before removing the frame.

After the frame is removed, the main PCB should be fairly easy to remove after unplugging the ribbon cables from the display and keyboard. Flip the chassis over and remove the display by firmly pulling it away from the metal backing plate, it is only held on with an adhesive pad. At this stage I got the second 8800 to the same stage, and reassembled my original phone around the chassis with the working keyboard mounted to it. It’s always worth checking that it is functioning before putting it back in its casing – You don’t want to put it all back together only to find there is a bad connection somewhere!

After doing a bit of research, this appears to be a fairly common issue. One suggestion that did work for me was to remove the bluez-* packages, but apparently there is the side effect of other Bluetooth devices no longer working. This isn’t an issue for me, so it’ll keep me happy until a real fix is in place.

Drop the following into your CLI if you’re having the same issue…
sudo apt-get remove bluez-cups bluez-pcmcia-support bluez-pin bluez-utils