Richard Leggett on New Media

It appears a new Views module has been released which is causing a couple of problems with the Drupal/Flash stuff, so I’ve updated the post to describe the necessary steps. Hopefully the Adobe DevNet article will also be updated soon with these changes.

You can read the updated post here.

I’m pleased to see that an article I wrote for Adobe DevNet has just gone live. “Creating Drupal Sites for Flash or Flex” describes the benefits of using a CMS for your Flash or Flex sites, how to set everything up, getting Flash talking to Drupal, and also covers the various modules available to power your site.

Drupal is probably the most popular open source CMS out there, it has an enormous user community, and the best thing of all is probably the fact that you don’t have to write any PHP or SQL to take advantage of this as a result! (Unless of course you want to, Drupal is fully extensible).

There are many ways to use Drupal with the Flash platform, ranging from simple content access, to automatically generated content that is formatted for and consumed by both Flash and non-Flash enabled devices. Without going into any more detail, please feel free to head over to the DevNet article to read more and download the source code and examples.

Update!

It appears there have been some name changes in the Services module called by Flash ("views.getView” becomes “view.get” and “node.load” becomes “node.get") - thanks to Stan for pointing this out in the comments. As a result I’ve modified the code and provided two versions below, the download I linked-to above has also been updated. The example FLAs are also now Flash CS3 compatible.

• Download source/examples (for Services 6.x-0.11 or newer, uses views.get and node.get).
• Download older version here (for Services 6.x-010 or older, uses views.getView and node.load).

Update 2!

It seems the new Services module also has a breaking change when it comes to generating the sitemap, you’ll see no properties on your nodes except id. Do not download the official 6.x-0.14 release, instead get the 6.x-2.x-dev development snapshot as the bug has been patched (unless the official release is newer in which case the patch will already be in). For a deeper explanation, it appears the PHP was not loading the nodes you were requesting, so the only info it could return was the ID which wasn’t a whole lot of use for a sitemap!

This afternoon I caught a tweet from @afovea which said “Why is a floppy disc icon still indicative of ’save’ … when was the last time you used a floppy disc?”. He has a very good point, we are stuck with quite a few analogies which have since become anachronisms; no longer part of how we physically use hardware or software; even “files” and “folders” are struggling to migrate the generations, splinted by Spotlight and Windows Search. Why is it so hard to think up an alternative? I think one reason is that “saving” itself is an outdated concept that might need re-thinking as we become fully connected, all the time.

One way of thinking of a “save” is the operation you perform when you want to package up a discrete version of something; to snapshot something in time so that it can later be retrieved. In some cases saving is not required, any email/webmail app for example, the save option here is “save as” which outputs what you are looking at into an archived format. This is generally used where forwarding will not suffice and search is lacking (with the exception of Gmail this includes most email clients).

I started to look at the software I regularly use and realised about a half has a “save” button/menu option, I included things like Facebook and Twitter (web) as well as Charles, OpenOffice and Flex Builder (desktop). Almost all software used to have a save button; you’d be editing a document or playing a game and you’d save your document or progress to disk, to hard drive, maybe even to USB. Along with save we had export, another way to get a document or file out of an application in another format. Then came save for web, a more specific version of export which saved a document in a web-friendly format for others to see.

RIAs

Then something changed. RIAs got serious, and after the shopping carts came fully featured document editors like Buzzword, Google Docs, Photoshop Express. Of course most interaction on the web was —and you could argue still is— not document based, it was/is purely passive. When RIAs brought editing to the browser, they also brought with it the heart of the web, collaboration. Editing and collaboration infects all passive media given time.

In order to allow collaboration, the best of these RIAs adopted a source-control-like document “history". No discrete files ever exist in reality, just markers for “this is how your document looked two weeks ago, or before Bob edited it 3 seconds ago". This new arrangement allowed real time collaboration, supported a less committed approach to editing documents, and with everything being on-line, allowed new live elements and interaction between documents previously impossible in desktop software (see the Aviary suite of tools).

So we’re now at a stage where we still have a “save” button in the aforementioned apps, but what is this really for? Is it a legacy mindset support? Is it to satisfy the user that they’ve just labelled a version of a document as “done for now". Is it because internet connection is still not 100% rock-solid and high-speed so that we can guarantee you are always “synced"; that simply closing or crashing the browser/AIR App/Silverlight OoB App doesn’t result in you losing your recent changes?

But what about… ?

In thinking about this I realised there were still good, legitimate reasons for having the concept of saving, right now. One argument for saving is that it allows someone to edit a document/entity and then abort, losing the changes. This is true on the whole, but an automatic history handles this. Another is the “save as” feature, allowing a copy of a document, a draft, a different version. I concede to this, but I think an updated paradigm for creating document-based apps could circumvent this.

Funnily enough, the AIR application I’m currently working on is document based, but we’ve found there’s no need for a save button. It happens automatically as you edit/navigate the app. Of course I had to build in undo/redo before this was reasonable.

The end of save?

I see the save button going away in most consumer applications as they migrate to more portable, inherently connected RIAs. Perhaps it’ll stick around longer for the pro-apps, the really heavy hitters, but even those in time will go. So what do you think? Is save going away, is it here to stay? How many apps do you use that actually require a save operation? What are you saving? Whatever the answer, for now at least I have to hit “save” so that you can read this post.

The Monkey Island games were something of a cult phenomenon, certainly back when I was at school, even though the first two were already retro at that time, I think that just added to their charm. Easily one of the best point-and-click adventures of all time.

The Secret of Monkey Island and The Curse of Monkey Island: Le Chuck’s Revenge were VGA masterpieces. The story and humor absolutely first class. The later upgraded 2D and then 3D sequels still hold a candle to the originals, but the news is there’s a new episodic Monkey Island coming out, and a remake of the original to boot.

So if you haven’t enjoyed the satirical humor of the first Monkey Island, here’s your chance to in new-fangled, more-than-10-color, super-mega-vga style without having to resort to the ScummVM emulator.

It’s fair to say I’m barely a gamer now, but I’m off to pre-order this one…

http://www.worldofmi.com/

I was up extremely late last night figuring this out, so I made some rough notes as I went along. Hopefully they’ll save someone a small nightmare.

Disclaimer: I haven’t written C code since I was in my teens, and apart from some Objective-C from the comfort of XCode, I haven’t really done a whole lot of GCC compiling on any system, so this may not be the best way to do any of this.

Flash Video

First of all, what are we doing? Compiling a portable FFMPEG binary from source so that we can include the exact libraries we want and run FFMPEG from another application. Why? In short, I want to convert almost any video format to… FLV/F4V. Technically the extension is irrelevant, what we are talking about here is the video codec used, which is either the Sorenson H.263 codec (Flash Player 6+), On2’s VP6 (FP 8), or H.264 (FP 9 update 3).

The F4V extension is really for files that use the H.264 codec in an MP4 container. So it’s up to you which you target, but it will impact on what you build into FFMPEG. The audio-codec used is generally MP3, but for the H.264 video it is AAC or MP3 (AAC is used in most MP4 files, such as podcasts from iTunes).

Now when you convert a video using FFMPEG, it will figure out what format to use based on the output file extension (e.g. flv), but you can supply extra parameters to choose a specific audio or video encoder.

For more info on the codecs inside FLV files see this Wikipedia page.

Compiling C-based Projects from Source on Mac OS X

Below are the general steps to compiling C-based projects on your Mac.

• Install XCode (developer.apple.com) - this installs the required GCC compiler.
• Download and extract your source code, for example, the lame library.
• Navigate to the extracted folder in Terminal.
• Run ‘./configure –enable-static –disable-shared‘. This will set up the make files with the given options for compilation, these options depend on the project, but they can be as general as where to find linked libraries, or what features to enable in the end product. In this case we’re doing a standard static compilation (this means a user won’t require shared libraries on another computer as we will compile them in and link to them statically).
• Run ‘make && sudo make install‘. Here I’m “making” aka compile and linking (you’ll see the binary appear in the source folder), then calling sudo to allow super user permissions and finally I’m also “make installing” it -copying it to the relevant place on my system so that I can invoke the new functionality from anywhere, and copying the libraries to my /usr/local/lib folder so that I may use them in other projects… such as FFMPEG!

Note: If you used something like Macports (or Fink) to install lame and any other video or audio libraries in the past, you’ll find that it installed them as dynamic libraries, dynlibs (and probably put them in /opt/local or /sw). This won’t do as we cannot statically link to these when compiling FFMPEG, we require the “.a” files. So I’m afraid that means you have to download the source for these separately, and when you ./configure them, make sure you specify –enable-static. You should then find them located in /usr/local/lib (by default) so that the linker can pick them up when we build FFMPEG.

I would take some time to look at the difference between static and dynamic libraries in *nix environments if like me this sort of thing is either forgotten or way over your head:

http://www.yolinux.com/TUTORIALS/LibraryArchives-StaticAndDynamic.html

Compiling the Dependencies

Download the source for the desired extra libraries, such as lame (http://lame.sourceforge.net/), x264 (http://www.videolan.org/developers/x264.html), faad and faac (http://www.audiocoding.com/downloads.html). You can then extract each to a folder, and follow the above instructions on Compiling C Source on Mac OS X. With x264 I had trouble building from the provided tar files, so I used git to get the latest source from their repo, but that requires you have compiled and installed git and Yasm (done in exactly the same way as all the rest of these projects).

Tip: If you download FAAD and FAAC, make sure you download the “bootstrapped” source as that includes the configure tool.

Compiling FFMPEG

When you compile FFMPEG you have two choices as to how you compile it, –enable-static, or –disable-static. This option dictates whether the resulting binary (app) includes within it any dependent libraries, such as libmp3lame used in MP3 encoding/decoding and FAAD for AAC audio decoding (for example in MP4 files). In this case I’m not bothering with FAAC as I am not encoding AAC, but I do want to statically link to libmp3lame.

When you go to compile FFMPEG, follow these instructions, and all should go well, giving you an ffmpeg binary that has everything it needs to convert video with the formats and codecs mentioned.

http://stephenjungels.com/jungels.net/articles/ffmpeg-howto.html

Using FFMPEG

Here’s the command line required to turn a sample video into an FLV file using MP3 audio and H.264 video codecs (I know, I know, I should be using MP4/F4V but I had serious problems getting that to play in anything). This file should play in any well-written video player as long as you require Flash Player 9.0.115.0 or higher.

Bash:

You can probably guess some of the flags, -y = overwrite file, -acodec = audiocodec, -vcodec = video codec, -ab = audio bitrate, -ar = audio sample rate, -b = bitrate, -g = frame grouping (used for keyframe creation/scrubbing).

Missing MetaData

So it may create a video that you can play, but you can’t see the video’s duration or actively scrub to another part of the video (I recommend installing an offline player like Wimpy FLV to test your videos easily). This is because FFMPEG does not inject the metadata into FLV files. But there is a cross-platform GPL’d tool that works (and they supply pre-compiled binaries!)

http://code.google.com/p/flvmeta/

Simple run ./flvmeta input.flv output.flv to inject correct meta-data.

The coders of these projects are true heroes of the community. Thank you.

Wrapping up

In my case I did not specify the –enable-non-free flag when ./configure’ing FFMPEG so that I can redistribute the binary, this meant until the FFMPEG team make their own AAC library, I cannot enable libfaad and libfaac to perform AAC encoding/decoding for my H.264 videos (so I’m sticking to MP3). Apparently some faac code has conflicting non-GPL licenses carried over from code written at companies.

Also remember you have to adhere to the LGPL or GPL license when including FFMPEG with your app. You also have to consider extra patent licenses for the codecs used, specifically the MPEG-4 codecs require licensing from the MPEG-LA, and MP3 from Thompson. The cost of licensing is apparently very low. The details of what licenses you require are a horribly grey area, it would take a specialist lawyer to really figure it out, if they could at all, and it depends on what country you are in anyway (see the FFMPEG homepage for more on this issue).

This same approach works for Windows however you’d be using the MingW compiler and probably want to have Cygwin installed to give you a *nixy style setup so the above instructions still work. However it’s much easier to find pre-compiled executables for Windows, in my case I found ones that included everything I had included in the Mac build so luckily I avoided this.