Two approaches to Flash [LWN.net]

The free software community has two independent projects working toward the implementation of a free Flash player: Gnash and swfdec. There has been some talk recently about these two projects, their goals, their accomplishments, and whether it makes sense to have them both. In an effort to bring more light to the situation, LWN held a conversation with the principal developers of both projects.

LWN: Could you please introduce your project: its goal, what you think are its major strengths, and what you would most like to improve?

Benjamin Otte: The obvious goal of Swfdec is to create an open flash player that is acceptably similar to Adobe's flash player. This means that the average end user will install a Linux distro and automatically get Flash functionality in their web browser, and never know or worry about installing an add-on component.

Of course, attaining this goal will open up so many opportunities that one might consider the opening of these opportunities as the real goal. Some of these are concrete -- improving accessibility of Flash-based web applications, tighter integration with the desktop, automatic blocking of advertisements. Others are more nebulous -- now that you have a feature-rich, graphics-oriented managed platform, what can you do with it.

Rob Savoye We are the "GNU Flash" player, Gnash. Our goal is a multi-media architecture for a networked world that happens to be Flash programmable. Gnash is the client side for streaming multi-media. We also have a Flash media server project underway, and a large database of meta-data and media content so people can find things better than using Google. Gnash is the only currently released component, the rest will be announced over the next few months. All are GPL'd projects, and support multiple platforms.

Our major strengths are portability, and real support for small embedded systems. Gnash currently supports all the BSD and GNU/Linux distributions, as well as 64 bit systems, and embedded systems like the OpenEmbedded, and the OLPC. Gnash also runs on WinDoze and Darwin as well. We also support creating custom extensions to Flash, so it's now possible to embed a device driver for lets say, hardware buttons on your smartphone, to trigger event handlers within a Flash movie.

I'd like to improve the completeness of our ActionScript support, there are some classes that exist only as stubs. That and adding Flash v9 support. This is all on the roadmap, but we need more developers to make better progress.

LWN: How would you describe the current development state of your project?

Rob: We're in pretty good shape, although Gnash is not a fully complete Flash player yet. The video support is progressing nicely, and we will have full support for Lulu.tv, YouTube, MySpace, etc... in the next release. Streaming video works in CVS, now we're working our way through the obscure bugs in the various video players to have this work 100% correctly. After the next release, we'll be focusing on performance issues, and finally starting to look at Flash v9 compliance.

Benjamin: Swfdec is actively going forward. With every release it supports lots of new Flash files, and they currently come out circa monthly. But as I mentioned on my blog it's hard to determine how far we've come as a complete Flash replacement. It's certainly still closer to the start than being a done.

LWN: Tell us about your development community: how many active contributors do you have? Why should developers join your project?

Benjamin: These days the project is primarily run by me. David Schleef, the previous maintainer is the other person that hacks on the Swfdec core. But Swfdec integrates into the greater Freedesktop and GNOME world, so we can draw from the knowledge and input of hundreds of developers. I'd like to highlight in particular the Cairo or GStreamer people, which are invaluable in making Swfdec work. And then there's a pretty big community building around Swfdec. Packagers, bug fixers, people that want to use libswfdec in their applications or those that contribute bindings to other languages like Python.

And developers should of course join the project because they want to and hacking on Flash is fun. For me it's not just wanting to play Flash files, but it's also that Flash is a platform. You can work on anything: a script language interpreter, rendering, multimedia, accessibility or security. And code that you write is immediately leveraged -- you make one thing work, and 100s of flash files start working.

Rob: We have about 6 core developers right now, and are looking for more of course. What we offer to developers is several things. One, we are a GPL'd project, which is important to some open source developers. We also have funding now to work on Gnash, which is also useful for some people to pay their bills. :-)

LWN: A working free Flash implementation would seem to be a high priority for the more desktop-oriented distributions. What sort of support are you getting from those distributors? Why do you think it's not stronger?

Rob: We get zero support from any of the distributions, we'd love to see that change. I occasionally hear from the package maintainers about packaging issues, but that's about it...

Benjamin: It would seem that way, but it isn't. For example, a free Flash implementation was not on Ubuntu's list of priorities at the last Ubuntu Developer's Summit. It seems that a lot of distributions are content with shipping solutions that are closed software when no Free software exists for a purpose. Another example for this would be Free drivers for modern graphics cards.

But I don't think it's solely the distributions' job to point out and work on deficiencies in the Free software world. It is also the job of the community. Distributions often just package what the community makes available to them. But it seems the community has also has been content with the closed solution.

LWN: Some LWN readers have complained that having two projects aimed at implementing Flash is divisive and wasteful. How would you respond to those readers?

Benjamin: The optimal number of projects for a given project space sounds like a good PhD thesis topic. Having multiple projects in a space, or multiple solutions to a problem is simply how things work in the community. Any non-trivial bug or project space has multiple solutions, and often one cannot determine which is the best solution until all have been tried. Also, people working on these projects are real people with real interests and complex motivations for working on particular projects. Simplifying it into "you currently work on A, so you'd instead like working on B in the same project space" is unrealistic. And IMO, divisiveness between similar projects often has more to do with fanboys than it has to do with developers, who obviously share interests and experiences.

Rob:Maybe in the proprietary world this would be a problem, but not in the open source world when we are both reverse engineering how Flash works. It's easy to get into tunnel vision with a single implementation, as some are bugs in our code, and others are bugs in how we think Flash works internally. Having two projects with very different implementations is very useful in finding the differences between our bugs, and Flash internals issues.

There is also a lot of coordination on things like test cases, specifications, and documentation that are shared between the two projects. So why is this a problem for some people is beyond me...

LWN: What else do you wish I had asked? Answer too, please :)

Rob: The Adobe EULA for Flash forbids anyone who has installed their Flash tools or plugin from working on Flash technologies. This has had a chilling effect on the development of free Flash players, since a developer must either choose to decide that Adobe won't sue them over this, or to do what Gnash does, which is a slow and inefficient, clean room, reverse engineering project.

Adobe has declined to comment on this issue, since the confusion benefits their lockin of the market. Although Adobe has said they support Open Source projects, and donated Tamarin to Mozilla, we'd love to see a public statement that Gnash developers won't be subject to a lawsuit. It's very difficult to find developers that have never installed the Adobe software ever, which is what we've been doing to maintain our clean room approach.

Benjamin: A lot of people would describe Flash as an "evil" format. What do you think the Free software community can gain from Flash other than supporting yet another closed format?

Flash provides features that are unmatched in the current Free software world, even when comparing it to the newest projects in the realm like SVG. And Flash is a pretty simple format. People have done great-looking sites, games or applications with Flash in your browser that are currently hard or impossible to achieve on any desktop. And they did that 5 years ago. So even if the Free software world were to not use Flash it certainly can learn a lot from it.

Another often overlooked thing is that there are a lot of content creators that create new multimedia content daily. Other people like Lawrence Lessig have already said that it is important to introduce them to Free content. In providing Free solutions to those artists and developers, the Free software community has a lot of mindshare to gain.

(Log in to post comments)

Two approaches to Flash

Posted Apr 4, 2007 18:01 UTC (Wed) by proski (subscriber, #104) [Link]

I think two assumptions in your comment are wrong. One is that it's possible to find out if a plugin has failed without human intervention. The other is that the user cares about the flash embedded in the page enough to be bothered by questions.

Two approaches to Flash

Posted Apr 4, 2007 18:19 UTC (Wed) by JoeBuck (subscriber, #2330) [Link]

A severe error could be caught, but if the problem is just that the plugin is scribbling all over the screen and making horrible noises, there iesn't a way to tell.

I suppose there could be a "meta-plugin" that would ask the user which plugin to use, and whether to ask the question every time, or to always use a given plugin for a given site. Either the user would answer, or the answer would be obtained from a database, and then the requested plugin would be invoked.

So if it turns out that FooFlash is better for YouTube and BarFlash is better for Pandora, the correct plugin could be invoked for both (this could be used for folks who cannot or do not wish to use Adobe's proprietary plugin).

Two approaches to Flash

Posted Apr 4, 2007 18:31 UTC (Wed) by elanthis (subscriber, #6227) [Link]

There's no advantage to this though. If you're going to have the non-Free flash plugin installed anyway, why even both with a Free one? The whole point of having a Free plugin is to not require the non-Free one to be installed. Why ask the user, "do you want something broken or do you want something that works?"

Unless the Free plugin is going to do something that the non-Free one doesn't, and the non-Free one continues to work on stuff that the Free one doesn't, there is absolutely no reason at all to ask the user a question about which to use.

Normal users will continue to use the non-Free one, FOSS sympathizers will use the Free ones, and this will continue until the Free ones are good enough that they can replace the non-Free plugin for just about everybody.

Two approaches to Flash

Posted Apr 4, 2007 19:37 UTC (Wed) by bronson (subscriber, #4806) [Link]

I don't think this is true... Most of my Windows-using friends use Firefox as their primary browser. If a site renders wrong or video streaming is broken, they temporarily launch IE, work around the problem, and go right back to browsing in FF. Why do you think it has to be all-or-nothing?

Two approaches to Flash

Posted Apr 5, 2007 1:17 UTC (Thu) by elanthis (subscriber, #6227) [Link]

You are lucky in your friends, then. The vast majority of regular people who have no irrational, social, or philosophical attachment to a technology are going to (maybe) try Firefox, find out it doesn't work for some sites, and go back to IE.

In all honesty, for those very people, Firefox is useless. It does nothing truly remarkable that IE does not, and all it does is add yet more semantic overhead in that it requires the user to identify a problem and work around it by using a second solution that would have worked perfectly 100% of the time for all their browsing needs.

Sure, IE may be a poor choice for some sites. There are even sites and web apps that work only in non-IE browsers. I've written more than a few myself for intranet usage and the like.

In the end, you can at best say that users have a choice between two different varieties of broken. Instead of fostering that kind of lunacy by forcing users to make choices, the better idea is to just improve one of the choices to make it 100% usable for everyone. Whether that choice is Firefox or IE is irrelevant to most people; they just want something that works.

For the goal of pushing Free Software, you'll never win by simply making a "good enough" clone of a popular product. You'll get the geeks on the bandwagon, and that's it. Some of those geeks will try to push their ideology on their friends and family - something I tried to do once upon a time - but that behavior is a complete disservice to the friends and family. If, for example, my mother is completely happy on Windows, with AOL, and IE, and Microsoft Office, and there is nothing lacking from her computing experience, she has nothing to gain from a practical and realistic standpoint. She doesn't care about FOSS, and she doesn't have to. She also knows from experience that she prefers AOL over any other portal/mail providor, that Firefox does not work on some of the sites she visits, that OpenOffice cannot handle all of the documents she wants to use, and that Linux cannot run some of the weird niche apps and games she likes to use.

Your approach might be to force her to run Linux, with a Windows dual boot, and force her to restart her computer or start up a virtual machine to run Windows whenever she needs those little extras. But she has gained nothing. She has lost. She now has to do more work to get all the functionality and use she had before, and she has gained no practical benefit from any of it. The Linux install does nothing that she needs that Windows didn't do. She can't even claim to have saved money or become "ethically pure" by using FOSS. All she gets is a pain in the ass.

Same goes with Flash. There is no huge benefit to running Gnash or swfdec over the Adobe plugin right now. This may not always be the case - the Adobe plugin is unstable in my experience, so Gnash and swfdec might one day be more reliable than Adobe's plugin. It will be nice if distros can come with Flash functionality pre-installed. At some point, it might be that installing the Adobe plugin is the action which decreases functionality or increases semantic load for no gain.

Right now, that's not the case. For the average user, the Adobe plugin does everything, and the Free alternatives add no value. They're interesting only to FOSS advocates. And even then, there's no benefit to a FOSS purist to be able to choose between "broken" and "evil."

The only people who have any practical need for such a plugin-selection feature are the early adopters and beta-testers, who I dare say are very small in number - too small to warrant Mozilla adding support for runtime plugin selection.

If someone decides to write a plugin wrapper, more power to them. I won't use it, but I have no reason to stop someone from writing such a thing. Their time, their choice. However, I will complain the second someone tries to force such a thing into distros, decreasing the simplicity and increasing the barrier to use solely for the benefit of a small handful of people who care neither for true practically nor for true Freedom.

Two approaches to Flash

Posted Apr 5, 2007 3:02 UTC (Thu) by xanni (subscriber, #361) [Link]

"There is no huge benefit to running Gnash or swfdec over the Adobe plugin right now."

There is at least one: The Adobe plugin is not available for PowerPC Linux, as used on older Macs. My wife has to use MacOnLinux every day to view Flash websites including Youtube.

Two approaches to Flash

Posted Apr 5, 2007 13:48 UTC (Thu) by dark (subscriber, #8483) [Link]

The only people who have any practical need for such a plugin-selection feature are the early adopters and beta-testers, who I dare say are very small in number

But those are the very people you need to attract in the early stages of a project.

Firefox is useless?

Posted Apr 5, 2007 15:11 UTC (Thu) by phip (subscriber, #1715) [Link]

> In all honesty, for those very people, Firefox is useless.
> It does nothing truly remarkable that IE does not,

Firefox does tabbed browsing, popup blocking, warning the
user before executing downloaded software, and it supports
add-ons that do gazillions of other usefull things. It
also runs on more platforms and operating systems than IE.

Sure, IE has added some of these features in version 7,
but there never would have been a version 7 without
Firefox to light someone's tail on fire at Micros~1.
Prior to version 7, IE had gone about 10 years without any
sign of innovation (unless you count the Google Toolbar...)

Firefox is useless?

Posted Apr 6, 2007 0:35 UTC (Fri) by giraffedata (subscriber, #1954) [Link]

In all honesty, for those very people, Firefox is useless. It does nothing truly remarkable that IE does not,
Firefox does tabbed browsing, popup blocking, warning the user before executing downloaded software, and it supports add-ons that do gazillions of other usefull things. It also runs on more platforms and operating systems than IE.

I believe the point is that for the people in question, none of these features are truly remarkable. And I agree. They don't want great; they want tolerable.

Sure, IE has added some of these features in version 7, but there never would have been a version 7 without Firefox ...

This is even less motivation than tabbed browsing, etc. for these simple users. They aren't going to switch browsers just to put pressure on Microsoft to improve its product. And neither will one of these users use Gnash just to send a message to Adobe.

Firefox is useless?

Posted Apr 7, 2007 0:11 UTC (Sat) by phip (subscriber, #1715) [Link]

> This is even less motivation than tabbed browsing, etc. for these
> simple users. They aren't going to switch browsers just to put
> pressure on Microsoft to improve its product. And neither will
> one of these users use Gnash just to send a message to Adobe.

I think you've got it backwards. Microsoft improved their product
because they saw their "simple users" switching to FireFox to get
features like tabbed browsing, safer handling of executables, etc.

Two approaches to Flash

Posted Apr 6, 2007 20:52 UTC (Fri) by im14u2c (subscriber, #5246) [Link]

There are more reasons to use Firefox than "irrational, social, or philosophical attachment." For me, it simply provides a significantly better browsing experience. Sure, every so often (and it's gotten exceedingly rare for me actually), I need to fire up IE for a site. Otherwise, I do everything in Firefox. Since all my bookmarks are here, my browsing experience is better under Firefox for a range of reasons, and it matches what I use under Linux, I think I have very rational reasons for not switching back to IE.

The same can be said about the free and non-free Flash plugins. I don't know if you've noticed, but ads and certain other content disable most of the options (such as "stop playing") in the context menu. That's actively hostile to users. There's no reason the free plugin should remove the user's ability to control it. But, if there's some site that has a Flash app that fails in the free plugin, then I don't see a reason not to have the non-free version as a fallback for that site.

And let's look at the dual-boot scenario. My wife runs Linux most of the time because it works well, doesn't crash randomly, and it works more like how she does. But, she'll reboot to Windows to play World of Warcraft. She doesn't stay there because she doesn't particularly like how Windows works. It's just a game launcher to her. It's not much different than having "game boot disks" 10-15 years ago... I don't think she's "lost" and she doesn't feel that way either.

Two approaches to Flash

Posted Apr 4, 2007 19:51 UTC (Wed) by ncm (subscriber, #165) [Link]

Elanthis has perhaps not noticed that there are two Free projects?

Two approaches to Flash

Posted Apr 4, 2007 21:16 UTC (Wed) by mrshiny (subscriber, #4266) [Link]

The problem is, asking the user which implementation to use is a sign of usability failure. For an app like Flash, there should be no question, ever, of this sort: the flash view should work, period. Frankly, the user isn't likely to even recognize the program choices:

Adobe Flash Player
Gnash
SWFDEC

As someone mentioned, if you have option 1 installed there's no need to ever use 2 or 3. And "Gnash" and "swfdec" have to rank up there with the worst, least useful project names ever. At least Gnash rhymes with Flash, so there is a mnemonic to let the user remember what it is, after they've had it explained to them. But the user will just read the dialog like this:

This page requires blah blah.  Blah blah blah:
[O]  blah blah
[O]  blah blah
[ ] Always do this from now on
OK - Cancel

Advanced users, who understand the situation better, may find this feature useful, but then the second problem is that needing to use two different tools to do the same job is dumb; it means the tools are broken.

[This post is not intended as a comment on the completion of the flash players. I can't wait until Flash works properly again, and we have the freedom with Flash that we've become accustomed to in other applications]

Two approaches to Flash

Posted Apr 5, 2007 4:50 UTC (Thu) by k8to (subscriber, #15413) [Link]

Sure, it represents a failure. It's a thought on how one might attempt to be (less) useful despite the failure.

The only downside I see is that in the time between semi-usable free flash and totally usable free flash (hopefully this window is finite), you have the possibility of people thinking this situation is acceptable somehow and making decisions that rely on that.

Two approaches to Flash

Posted Apr 5, 2007 10:38 UTC (Thu) by macc (subscriber, #510) [Link]

User interaction with the original player is severly broken
in all version i had the nonpleasure to test ( and in a
variety of browers).
The player window seems to take a global grab on occasion
and is then unresponsive for 60 seconds or more.

This is simply not acceptible.

As it is, Acrobat Reader has similar problems(and this lets me
think that commercial programmers can't jump ).

Two approaches to Flash

Posted Apr 7, 2007 10:59 UTC (Sat) by filipjoelsson (subscriber, #2622) [Link]

I use a FF plugin called flashblock. It replaces flash with a playbutton - and so my browser won't waste resources nor my attention with playing stupid adds all the time.

I suppose this plugin could be hacked up to deliver choice of which flash-plugin to use for flash playback (ie give me two or three buttons instead of just the one).