UPDATE 2010.12.21: DelicousBackup has been replaced by Yummy!. Check it out at http://www.bram.us/2010/12/21/introducing-yummy-a-self-hosted-delicious/. The link below won’t work anymore.

Ergo I wrote a little PHP app does exactly that: After having created a backup of your bookmarks import.php will import your bookmarks into a local database. Once imported index.php can take over and display them for you.

DeliciousBackup in action.

Note that DeliciousBackup is totally unstyled (for now) and only allows you to display your bookmarks (all, or per tag). Now, there’s only left the task in writing an API endpoint so one add new links to it, right?

The source of DeliciousBackup can be found on GitHub. Feel free to fork off if you like.

▼ Download DeliciousBackup from GitHub

Oh: Running your own Delicious will be the new running your own url shortener. Mark my words

Last night I whipped up a tad of CSS to position the tab close buttons in Firefox 4 (currently in Beta) on the left hand side of the tab. The CSS needs to be placed into your userChrome.css to make it work (the code was tested with Firefox 4.0b7).

/*
* Do not remove the @namespace line -- it's required for correct functioning
*/
@namespace url("http://www.mozilla.org/keymaster/gatekeeper/there.is.only.xul"); /* set default namespace to XUL */
/* move favicon, throbber and text to the right so that the close button appears on the left */
.tabbrowser-tab .tab-icon-image {
  -moz-box-ordinal-group: 2 !important;
}
.tabbrowser-tab .tab-throbber {
  -moz-box-ordinal-group: 2 !important;
}
.tabbrowser-tab .tab-label {
  -moz-box-ordinal-group: 3 !important;
}
/* hide close button initially */
.tabbrowser-tab .tab-close-button {
  display: none !important;
}
/* show close + hide throbber & favicon on hover */
.tabbrowser-tab:hover .tab-close-button {
  display: block !important;
}
.tabbrowser-tab:hover .tab-icon-image {
  display: none !important;
}
.tabbrowser-tab:hover .tab-throbber {
  display: none !important;
}
/* app tabs should not behave like regular tabs, so let them stay normal (viz. don't show close on hover) */
.tabbrowser-tab[pinned]:hover .tab-close-button {
  display: none !important;
}
.tabbrowser-tab[pinned]:hover .tab-icon-image {
  display: block !important;
}

The CSS initially hides the close button of all tabs. When hovering over a tab, the favicon is replaced by the close tab button. The code also plays nice with Firefox’ App Tabs.

Normal Tab

Hovered Tab

Hovered App Tab

As the code is placed on GitHub, feel free to fork off!

Introducing the Plex Media Server

The thing that got me all excited was an architectural change within the app: in Plex/Nine the Media Server (viz. the code that scans your disks, detects your media and builds up the library) got separated from the Media Center itself. By this, the Media Center (plex.app) was “downgraded” to being a client application which communicates with the Media Server.

The scanner code which steers the Media Server is, just as in Plex/Eight, really awesome: just tell the app where your movies and where your tv series are located and … that’s it. The thing is pure magic: it’ll detect everything it finds, download the description, fanart, etc.

Plex/Nine Sneak Preview, demonstrating the Media Server
(note that this footage is from a pre-alpha version. Stuff has changed by now )

Enter the magic

Not convinced yet? Hold on: Imagine you have a Mac Mini running Plex/Nine. The Plex Media Server on the Mini will build up a library that plex.app can use and play back for you. Now, if you have a second Mac (say, a MacBook Pro) with Plex/Nine, that one too will have its own library.

The magic part? plex.app from the MacBook Pro will also be able to communicate with the Mini’s Media Server (and vice versa)!

Yes, that’s right: the Plex Media Server allows connection of multiple (remote) clients! Above that this all goes automagically in plex.app: Any plex.app instance on any Mac within your LAN will detect all Media Servers running, load up their libraries and allow you to play back all media found by those Plex Media Servers. And it gets even better: remote libraries are shown exactly the same way as if they are loaded up on a local plex.app instance: For example, if you’ve watched a movie on the Mini, the MacBook Pro will also mark it as watched.

This structural change has proven its use a few times by now here at home. Just this week my girlfriend was watching the television (on which the Mac Mini is hooked) when I just wanted to resume watching an episode of The Big Bang Theory. No worries there: I started Plex/Nine on my MacBook Pro, scrolled through the (remote!) library and was able to resume the episode as where I stopped watching it on the Mini.

Digging deeper

To those wanting to know the tech behind it: Plex Media Server sports a built-in web server, listening on port 32400. If you go to http://localhost:32400/, you’ll see an XML being returned. Each of the keys shown can be attached to the URL. http://localhost:32400/library/sections/1/all for example will show you all movies detected by the Plex Media Server (that is, if you have defined a movies section in the Media Server). If you dig a little deeper, you’ll eventually find the URLs pointing to the media files themselves

Outside the box

When thinking outside the box and going beyond the “Media Center on a computer, hooked to your television” dogma, one can think of many uses thanks to the introduction the separate Media Server. What if, for example, you could have your Breakthrough Internet Communications Device™ (aka iPhone) play nice with the Plex Media Server?

Those familiar with Plex/Nine know that the above is exactly what the Plex developers have done: along with releasing Plex/Nine, they also release a companion iPhone application which loads up your library, and lets you playback your media. Yes, that’s right, you can watch your favorite films right in your bed, on your iPhone.

Now, I’m not talking about something like iTunes’ Shared Library where you get a table with a list of names, but a feature complete client with full integration which shows you a thumbnail, the description, etc.

“Remember, remember the fifth of November” — Plex for iOS in action.

Two weeks after the Plex/Nine release, an external developer also released an iPad Plex Client, named KLEXi

KLEXi, Plex Client for iPad

How they’ve done it? Simple: they’re communicating with Plex Media Server’s built-in webserver

The possibilities are endless

The Plex developers even took it to the next level by announcing a partnership with LG Electronics: any LG Netcast™ enabled HDTV or Blu-Ray device will act as a Plex client and will be able to play back media found on any Plex Media Server within your network.

And just today a YouTube video caught my attention: someone actually managed to hack Apple’s AppleTV2 and let it communicate with Plex Media Servers located on your network.

As all other clients mentioned before, this one too communicates with Plex Media Server’s built-in webserver. If you happen to own an AppleTV2, you can follow these instructions to make it all happen:

In closing

In closing, I think the separation of the Media Server from the Media Center/Player is one of (*) the best moves the developers have done. Being a web developer, I cannot help noticing a striking resemblance with how we build (some, not all) web applications today: provide a data extraction layer (API) and build your app on top of that (cfr. Twitter, Gowalla, etc.).

Plex developers, I bow for thee!

(*) Another great move is the implementation of a metadata caching server on their end, but that’s a whole other story

What I like about Tweet Nest is that it sports detection for sites like TwitPic, Yfrog, img.ly, imgur, etc: if an image is detected in the tweet, the image will be shown inline next to the tweet:

Tweet Nest detects images, and shows ‘m inline

As Instagr.am is fairly new, Tweet Nest doesn’t support it. Yet. As Tweet Nest is written in PHP, it was no biggie for me to dive into the (rather cluttered) code and add support for Instagr.am to it. The code below will add Instagr.am detection for newly retrieved tweets: if an instagr.am link is found, it’ll extract the photo from it.

A code insertion need to be done in the file extensions/images.php. First, locate the following code snippet in the file (the closing curly bracket should be on line 46):

if($domain == "twitvid.com"){
    $imgs[$link] = "http://images.twitvid.com/" . $imgid . ".jpg";
}

After that, insert the following code:

if($domain == "instagr.am"){
    $html = (string) getURL($link);
    preg_match('/<meta property="og:image" content="[^"]+"\/>/i', $html, $matches);
    if (isset($matches[0]))
    {
        $imgs[$link] = substr($matches[0], 35, -3);
    }
}

preg_match("@/([a-z0-9]+).*@i", $path, $m);

preg_match("@/([a-z0-9\-]+).*@i", $path, $m);

Savvy users will see that the code above is quite a nasty thing: fetch the Instagr.am page and extract the image from the HTML. Yes, scraping. Urgh.

I hope to change the code above to something nicer any time soon but that can only happen when Instagr.am adds support to having simple (redirect) URLs to fetch the thumbnails, just as TwitPic, Yfrog and many other services offer. If you feel my pain, please vote up the corresponding issue on the Instagr.am Getsatisfaction page.

Anywho, if all goes well, the result should be something like this:

TweetNest, with Instagr.am integration

Please note that this fix will only work for newly retrieved tweets, and not for the current ones imported (unless one tells me a way to re-parse all previously fetched tweets again)

That’s it … for now

(*) Actually not all tweets were backed up. The first fetch only got 3200 tweets, as limited by Twitter itself.

Gowalla Missing Items Location Finder — Now with OAuth!

Main change is the switch to OAuth. This way one doesn’t have to disable his/her Gowalla Passport Privacy anymore. As a result it is not longer possible to “spy” on an other user/username by entering his/her username.

Implementing OAuth was a walk in the park actually, thanks to the excellent documentation available. However, I still found some little quirks in it, and suggested the Gowalla team to improving the docs in some sections.

Next to the OAuth implementation, some of the screens/copy were revised/simplified, and the ability to automagically detect your location, via your browser landed into Gowalla MILF. One such a screen is the results page.

Gowalla Missing Items Location Finder — Revised Results Page

Newer updates will be spread via the Gowalla MILF Twitter Account. Perhaps a changelog will make it onto the site.

So, what are you waiting for? Go try out the improved Gowalla Missing Items Location Finder!

Gowalla Missing Items Location Finder — Welcome Screen

Development

The project was built in an afternoon, excluding the time fiddling with the (basic) Gowalla API and excluding the styling. Under the hood of Gowalla MILF one can find PHP, Spoon Library, Google-Geocoding-Class along with some extra custom classes such as one which communicates with the Gowalla API. Other technologies used are Google Maps (V2) and jQuery.

Gowalla Missing Items Location Finder — The Results

A vast amount of time was spent on minimizing the load on both my server and the Gowalla (API) server:

• Templates are pre-compiled, which is done by Spoon itself.
• All calls made to the API are cached into a database for two hours. This to prevent hitting a rate limit (is there any?)
• The coordinates used for fetching of the locations/items near a lat-lng pair are rounded to three digits, resulting in the same API calls for two sets of coordinates which are very close to eachother (and resulting in more cache hits).
• The transparent images used – which are fetched from Gowalla and then processed by PHP – are cached on disk, taking some load off my server when the same image is requested over and over.

For now the tool does not use Gowalla’s OAuth capabilities as I wanted to keep it simple to implement. However, OAuth might make into it as it would no long require one to have Gowalla Passport Security set to off whilst fechting his/her user info. Gowalla MILF now sports OAuth, read the announcement here.

Design

The design — if you can call that a design — is pretty straight forward and is built around the main map which one can see. Below are some extra screenshots of some special scenarios that might appear (no location, no items, no interesting spots or no spots at all)

Gowalla Missing Items Location Finder — No location (hometown not set)

Gowalla Missing Items Location Finder — Passport Privacy (No items)

Gowalla Missing Items Location Finder — No loot (no interesting spots around here)

Gowalla Missing Items Location Finder — No dice (no spots around here)

The used icons in the sidebar were all found via iconfinder.net

Happy item hunting! Now, go check out Gowalla Missing Items Location Finder!