One Year Later: Adobe Abandons 64-bit Linux Again

By Caitlyn Martin
June 6, 2011 | Comments: 55

Just under a year ago I wrote about how Adobe had abandoned 64-bit Linux, at least temporarily. Linux users who chose to run a 64-bit OS were left with a range of unsatisfactory choices: use an outdated beta with known security vulnerabilities; run an FOSS alternative, most likely gnash, despite limits in functionality and compatibility; or run a 32-bit browser in a 64-bit operating system. At the time the move was surprising since reviews of the 64-bit beta, like this one by Steven J. Vaughan-Nichols of Computer World, were quite favorable.

Three months later on September 15, 2010, Adobe announced a preview of Flash Player "Square", development code for an upcoming release of a native 64-bit version of Flash Player 10.2. Unfortunately Preview 3, released on November 10, 2010, was the last Square release. Flash Player 10.3 was released for 32-bit platforms only.

Once again there are known security vulnerabilities in the now eight month old beta and no patches are available. In addition, the community forum page for discussing Flash Player "Square" has moved and the link on the main page for "Square" is now broken. A reader (see comments below) found the new forum location on the Adobe Labs website. It seems the user community is complaining vociferously about precisely this issue: no security updates and silence from Adobe in response to their questions. If Adobe is continuing development on a 64-bit version of Flash Player they are not sharing any information with the public at this time. For the time being Adobe has effectively abandoned 64-bit Linux.

This decision makes even less sense than it did a year ago. 32-bit processors have effectively become legacy technology. Even the low end of the market, nettops and netbooks, now mainly ship with 64-bit processors. Yet it seems Adobe is unable to maintain a 64-bit Flash player for any platform: not for Windows, not for MacOS, and certainly not for Linux. Last year some of my readers questioned whether or not Adobe support for 64-bit operating systems, including Linux, is really important. I'll refer you to an excellent article by Stephen Shankland for CNet from last year which explains clearly why it is.

Is 64-bit Linux in particular important for Adobe? As much as some people would like to deny it, the Linux desktop market is not small anymore. Last year Bruce Byfield, in an article for Datamation, quoted KDE's Aaron Seigo placing Linux desktop adoption at anywhere between 8% and 12%. My own research and analysis lead me to write an article last September that also put the number at 8% or more. This is why Adobe has consistently offered a 32-bit Linux Flash Player.

Examining the Alternatives

The most common solution found in Linux community forums is to continue to run the outdated Flash Player "Square" Preview 3 and to use Flashblock to mitigate some of the security risk. Unfortunately this does not work as well in Firefox 4.x as it did in previous releases. To quote the Flashblock website:

Some flash is slipping past Flashblock on Firefox 4.0 (mostly Flash inside IFRAME subdocuments).
Even if this issue is resolved serious vulnerabilities remain whenever Flash is enabled.

The FOSS alternatives have improved since last year. I wrote about improvements in gnash 0.8.8 back in August. In March gnash 0.8.9 was released. The new version offers additional support and compatibility with current versions of Firefox, Chromium and Konqueror. However, gnash still does not offer full compatibility with SWF 8.x and 9.x and no support for features added in 10.x.

Lightspark is another FOSS Flash alternative under heavy development. It focuses on adding a limited set of additional advanced features on top of what is already included in gnash. As Lightspark falls back on gnash for basic functionality you ideally would want to install both.

The last alternative, running a 32-bit browser in a 64-bit OS, is certainly a less than elegant solution with technical limitations. In addition, a fair number of Linux distributions, including Slackware and most Slackware derivatives, do not include 32-bit compatibility libraries in their 64-bit builds. SalixOS, for example, is not installing any Flash support by default in their native 64-bit builds and offers gnash in their repository.

Will Flash Remain Relevant?

In response to last year's article the first time Adobe 64-bit Linux support disappeared, some readers expressed the opinion that Flash would become irrelevant as HTML5 adoption increased. With all major web browsers now supporting HTML5 I do expect that more web developers will migrate away from Flash. The lack of a native 64-bit Flash player may speed that migration. However, as long as sites people want or need to visit require Flash having a decent native 64-bit player will remain important.

UPDATE: I have received a response from Stefan Offerman at Adobe. Please see my follow-up article: Adobe: 64-bit Flash Player Later This Year.


You might also be interested in:

55 Comments

You can configure NoScript to work exactly like Flashblock with the added benefit of being reliable:
http://noscript.net/features#contentblocking
http://hackademix.net/2008/06/08/block-rick/

Still even with this, using outdated and buggy* Flash versions is a bad idea, IMHO.

(* there's famous bug report about a memcopy/memmove issue in Fedora's Bugzilla even Linus commented on)

What rubbish!! Flash Player "square" page on Adobe Labs is still there - open your eeyes

http://labs.adobe.com/technologies/flashplayer10/square/

Perhaps you need to read my article again. I never claimed "Square" wasn't there. What I wrote is that it has not been updated in eight months and has serious security vulnerabilities. I also made the point that there is no sign whatsoever that any development is in progress.

My article is accurate.

Looks like it's back to using nspluginwrapper and the 32bit plugin again for 64bit Linux users.

That never worked well for me. My choice during the last period of no support from Adobe was gnash. Thankfully gnash has improved quite a bit in the last year.

Well, I'm installing it by default for my 64bit Fuduntu users, so I am probably familiar with most of your issues. I have resolved every known issue, and it is working well for a few hundred of my users (or they just aren't reporting it being broken).

I would say that the best option for flash in a 64bit distribution is indeed nspluginwrapper and not downgrading to a 32bit browser, or using a plugin that's grossly out of date.

[root@host ~]# pacman -Qs adobe
local/flashplugin-prerelease 10.3.162.29-2
Adobe Flash Player Square

I wonder where Arch Linux found that because it sure isn't on the Adobe website anywhere. Can you enlighten us?

In any case, I won't consider Arch until they have signed packages. Trading one security nightmare for another potential security nightmare is not particularly helpful.

Fredrik's Arch package is from Nov 2010 and is from the Arch Linux User Repository. (AKA an old crufty version that hasn't seen a security update in >6 months? No thanks. Too long esp since it's Adobe.) The the Flash from the official Arch repositories is the 32-bit version + nspluginwrapper.

Also if you've tried nspluginwrapper before with limited success you shouldn't assume that will be the case for all time. The project is still pushing out new versions. The latest was a few days ago even! (It's not dead like Adobe at least.)

Doesn't look to me like it's been completely abandoned.

If you would have checked, you would realize that the Adobe forum is still up and running for Flash Player "Square"

http://forums.adobe.com/community/labs/flashplatformruntimes/square64bit?view=all

While it does seem that there haven't been any updates to the software in quite awhile, the forums are fairly active with most referring to problems and issues.

You think I write things without checking? Sorry, no. The link from the main Adobe Flash Player "Square" page which used to work is now broken. It seems that they moved the forum but did not update the main web page.

I read through the forums and it is entirely user/community comment. There is still no indication that any new development is taking place. It also seems Windows users are complaining about this as "Square" has issues with IE9. See: http://forums.adobe.com/thread/847238 Users in the forum are complaining about the silence from Adobe and the lack of security updates as well: http://forums.adobe.com/thread/837467

So, while I thank you for the updated forum link it seems the article, sadly, is still accurate. I really, really wish you had proven me wrong. That would have been better for everybody.

Thank you again for the link to the new forum location. I've updated the article accordingly.

Slackware 64 has a 32-bit compatibility layer from renowned slacker alienBOB! It's not installed by default but it's very easy to add.

I love the way the Slackware community tosses the word "easy" around and has a truly unique meaning for the word.

In any case, adding 32-bit support is non-trivial. It means replacing core Slackware packages with third party packages. The instructions are here. Savvy Slackware users will undoubtedly have no problem following them.

For those of us who want to stick with a 64-bit browser this is still not a good solution.

I've been a user of 64-bit Slackware-alikes for a few years now - started out on Slamd64, and ran that until it was abandoned with the release of Slackware64. I do run multilib for Wine and Skype, and probably will for years to come.

I remember with fondness the day the 64-bit flash alpha was announced - I remember throwing away nspluginwrapper and the 32-bit flash, and firing up 64-bit Firefox. It was a great feeling to be able to use Flash natively - it was an even better feeling when I realized, after some weeks of use, that I was having less problems with the 64-bit alpha than most others were having with the 32-bit full release, EVEN ON SLACKWARE itself. In general, the alpha crashed less often and played a wider range of content.

So, Imagine my surprise when the first beta came out - and it was even better! Through a period of weeks where I saw dozens of complaints about Flash instability in the ##slackware channel on Freenode, I saw almost no Flash-related crashes (the ones I saw I STILL can't avoid potentially putting them down to Firefox) and was able to access an even wider variety of online Flash-based content.

This brings me to today, where I am still using the 64-bit "beta" version. I'd argue that it's still just as stable as when it was released back in November. Naturally, it needs some security fixes - but I still get security updates on openssh and Apache, so I'm not terribly bothered by that. C'mon, Adobe, pay attention!

I will say, though, that even though I will maintain a multilib installation, there's no way I'm going back to nspluginwrapper. Ever. I'll stop using Flash before I put myself through that nonsense again.

Thanks for an insightful article, Caitlyn. Hope it gets some attention from those... err... "lovable knuckleheads" at Adobe.

But why would I want to run a 32 bit browser just because of flash? :-P

As an Arch user I was curious. So I looked. The flashplugin-prerelease 10.3.162.29-2, as seen above, is from AUR (Arch User Repository). A quick look in the PKGBUILD shows that it is actually 10.2 that is downloaded and installed!

See for yourself
https://aur.archlinux.org/packages.php?ID=32072


Very disturbing. Be careful with those unsigned user submitted packages.

Oh no, not this "We are 10%" drivel again...

Last time, some poor souls drank the Kool-Aid and started a Web site devoted to "proving" that desktop Linux was >1%.

http://www.dudalibre.com/gnulinuxcounter

And after nearly a year online, they have yet to reach even one-tenth of 1% (or anywhere close to it).

O'Reilly can and should do better than this.

Factchecker, you want me to drop all research and all market data and just blindly accept wildy inaccurate web counters? Sorry, no. The best estimates are between 8-10%. Since tablets are counted as part of the desktop market that number has likely increased since the article I wrote last year.

Perhaps, instead of insisting that O'Reilly "do better" you should drop your own biases and prejudices and do a little better yourself.

The aggregate of the major sites collecting statistics indicates an estimate of 1-4%, not 8-12%. None of the guestimates landing at 12% have ever made sense. I can see why a lot of people want the number to be 12%, but I believe we are artificially inflating it which will do us more harm than good if one of the major vendors decides to take "desktop Linux" seriously and the numbers just aren't there. It could cause a confidence problem that we as a community may never recover from.

Sales figures collected by the various market research groups for netbooks alone would give Linux at least 6% of the market. Another two percent for everything else seems reasonable. The web counters have been debunked time and again. The 1% - 4% number you quote is the one that is absolutely not credible to me.

I never claimed 12%, I quoted someone who claims it could be that high. Considering tablets running Android I could see us reaching that number this year easily. I don't think the numbers I claimed are inflated and I see more harm from deflating the numbers than in being realistic.

As you probably know ASUS is once again offering Linux (Ubuntu and MeeGo). As a company they have plenty of experience with just how big the Linux market is for netbooks and nettops and would not have reentered the market if they didn't think it was a lot more significant than the web counter numbers. Heck, it has been widely reported that they left because of strong arm tactics by Microsoft. Why take on MS if there isn't real money to be made?

I didn't mean to imply that you did claim 12%, sorry I meant to say 8-10% but 12 jumped out of my fingers and my mind. I see web stats as hard numbers, and discount claims of user agent strings because they aren't normally included with the browser so in my mind they carry more weight. If we are including Android in the statistics, then yes the number will definitely be higher. I'm hopeful, but remain highly skeptical.

I'm also familiar with the Asus / Canonical agreement. While it's good overall, it is a little painful for me personally since I don't think Ubuntu is the best product for that device. I am obviously biased since I have a Linux distribution that's designed to work well on those devices though.

Sales figures collected by the various market research groups for netbooks alone would give Linux at least 6% of the market. Another two percent for everything else seems reasonable. The web counters have been debunked time and again. The 1% - 4% number you quote is the one that is absolutely not credible to me.

Please don't confuse current sales figures with installed base. As for web counters being debunked, please show me where and how as I've never read a credible argument. A lot of unsubstantiated conspiracy theories, a lot of hand waving saying it's OK to count web counters for browsers but not for OS, etc. yes. But not one single solid, well defined argument explaining the flaws in the methodology for measuring installed base, not marketshare.

Or do you conflate the numbers for Android and Linux desktop? If so, you could make an argument that Linux PODs are now about 2-4% of the devices browsing the Internet.

(Again, don't quote the latest sales figures saying Android is outselling iOS and has been for a couple of quarters. I know that. The installed base for iOS is a lot larger, that's all. Remember, installed base, NOT current marketshare.

Web counters don't measure installed base. They measure people visiting web sites that pay to be counted and that is all they measure. When I was consulting for Red Hat I saw plenty of corporate desktops running Linux that have little or no exposure to the web. In my own office there are several machines but one common IP address (think IP masquerading) and rarely if ever more than one machine on the net at a time.

The only way to measure share is sales. I went with sales figures over a period of years from multiple sources. Is that 100% accurate? Probably not. It is a lot closer than web counters.

Android is a Linux distribution so, yes, it counts if it is installed on a desktop, notebook, netbook or tablet. It does not count as a desktop OS when installed on a phone. That is a different market.

Quoting someone else is NOT "research."

So, tell you what: Publish enough of your "research" methodology to allow someone without *your* biases to replicate your results.

Otherwise, you are just pulling numbers out of... um... thin air.

Of course compiling data from multiple studies is research. Scientific and medical journals publish that sort of research all the time. You, on the other hand, claim Linux has significantly less than 0.1% market share with one counter link... and you accuse me of pulling things out of thin air? Laughable.

Anyway, this whole argument is tangential to the article at hand. Please save further comments on market share until I write another article on that subject, which will happen when I have significant fresh data to report on.

There have been some issues with 32bit flash and 64bit distributions, but for the most part it works fine. It would be nice to one day see a production ready 64bit flash, but for now I'm OK with it remaining 32bit and using nspluginwrapper.

At several points, including Preview 3 of "Square", the native 64-bit player worked as well as any production player Adobe had ever released. That's why this is all so puzzling to me.

The future is 64-bit, not 32-bit, as the Shankland article makes clear. Getting a 64-bit client out there is going to be vital to Adobe if Flash is to remain relevant.

I am going to have to overcome my allergy to PulseAudio and try building Lightspark again for the distros I use. It should be interesting to see how well the gnash + Lightspark combo is doing at this point. The best solution, IMNSHO, is for the FOSS equivalents to do a good job and to leave the proprietary software vendor behind.

I think that's where we probably agree to disagree, I have no issue with software being closed source as long as it solves the problem it is intended to solve, and doesn't cause problems with other software packages surrounding it. :)

I'm sure the 64bit package worked well, I remember testing it. I don't have any issues with the 32bit package that haven't been corrected though.

Slightly off topic, other than on the general theme of companies involved in proprietary stuff covering their eyes whenever Linux comes up, but I notice the link in the September article you mention here saying "Dell is offering laptop and desktop models" now goes to a page with single a lonely-looking low-end desktop workstation, and my attempts to price out a real Linux laptop with them last week gave me confirmation that all they still offer are a couple of netbook models and one slow, grossly underpowered "Celeron" laptop.

There are a few others you see if you use their search function that look promising, but when you click on them you find "Ubuntu" is no longer an option for them.

Very irritating for someone like me looking for a full-power Linux laptop within a constrained budget. Fortunately there are a few real dedicated Linux-laptop companies surviving out there, so the market can't be all that bad (Ohava computers, System76, Frostbite systems, Zareason, Emperor Linux...), so I'm guessing Dell is still just using Linux for license-fee bargaining with Microsoft.

Back on topic: back when 64-bit Flash for Linux was first being publically tested, I seem to recall that the guy in charge of development appeared to spend much of his time complaining about how hard Linux is. I don't know if that's changed at all in the years since then, but I haven't had much faith in Adobe's interest or ability on Linux since then.

The rumour mill that is Phoronix had a story claiming they had a contact in Adobe who said they had been working on an updated 64 bit flash. Obviously care must be taken as Phoronix have made some big claims over the years which have yet to come to pass.

Additionally the Adobe Linux engineer we used to hear from was Mike but whatever he posted eventually attracted negative and abusive comments (even if topic was interesting) which probably took the fun out of posting technical stuff.

Adobe hasn't got a particularly good reputation for how it handles security. No matter if it's about supported software or not. In a sense it looks like Adobe products are best used off-line, which makes the widely use of Adobe flashplayer awkward.

Anybody using gnash? Is it ok?

I just started playing with it on my 64-bit desktop running SalixOS. I also want to build a package for Lightspark and try the two together. I will write about the results in a few weeks.

I had pretty good results with gnash 0.8.8 with two caveats: it can get pretty darned memory hungry and it doesn't work on all sites.

1. screw Adobe and their sycophants in `web designing'

2. I love the way you take no grief. Keep up the good work.

"KDE's Aaron Seigo..."

No credible article can quote Aaron. He writes with facts out his ass. 8-12%? Ha ha, Really? I am sure Mr.Aaron also thinks KDE runs on practically all computers. Please stop taking the KDE folks seriously.

Actually, Bruce Byfield quoted Aaron Seigo and I quoted Mr. Byfield's excellent article. Mr. Seigo made a very cogent point when he stated that Linux has much more penetration in China than it does in the U.S. or some European countries. That is at least partly due to the fact that the Chinese government has it's own distro, Red Flag Linux, and also promotes use of the Longsoon (64-bit MIPS) processors which are produced in that country. His point is valid and web counters on English or European language sites don't count the Chinese particularly well.

It's one thing to be dismissive. It is another again to make a cogent argument backed with facts. You'd be much more convincing if you tried the latter.

Really? Linux is making inroads in China?

It's not that I don't want to believe you but my limited experience tells me otherwise. Nothing scientific here but after spending quite some time in China Windows seems ubiquitous. A few examples:
- IE is said to be used by more than 90% of the Chinese internet users.
- A lot of them still use IE6 for fear of breaking compatibility with numerous sites.
- Last time I checked, online banking (at China Merchants Bank) required Windows and IE.
- The Green Dam software only worked on Windows and usually they're serious about censorship.
- Most if not all Chinese universities are said to use a authentication client named Ruijie Supplicant that is again Windows only. There is a fix now, though (look up mentohust).

When I'm in China I rather get the feeling that it's a Windows' world.

Cheers

Caitlyn Martin,

Great work. I second Lefty's comment; take no grief!

I hate adobe and flash. I hate the fact that I'm totally locked into their software for work. :(

Flash can't die soon enough.

Are you on twitter or identi.ca?

Well, I was very surprised that two different ways of counting GNUlinux penetration lead to the same results:

Caitlyn Martin took , on a world wide scale (not on my little French province town, neither in a limited bunch of friends sharing the same tastes) the number of PCs with GNUlinux preinstalled (never paid the Windows tax). This has some economical sense, and was validated by Windows being worried with GNUlinux increase.
However, it is based on a number of sales per year, and is criticised because there are PC which are older... (in physics, it would be like approximating position with speed).

The other method was to count the number of security updates of Firefox, and assuming
* 90% of the PCs were connected to the Internet (this percentage cannot, obviously, be deduced ... from browser statistics....)
* a large part (and unknown) of FF updates is through distribution's package manager (not *.tar.gz ), leading to an underestimation of GNUlinux penetration with this method..

This does not take into account PCs which, after paying the M$ tax (this is the only point of interest of Micro$$$$oft), switched to GNUlinux...

The two ways of countig have drawbacks, but these drawbacks are not the same : in physics, when two imperfect measures lead independently to the same result, the result is accepted.

On my 64-bit host, I just install Virtualbox and then create a 32-bit linux guest. It works well enough for me.

Why would anyone want to install Virtualbox and create a virtual machine just to run Flash? That's beyond overkill from my perspective.

Very few people know about the existence and functionalities of virtual machines... (any way, thanks for you as your idea gave me an hint of how to read an external USB disk, formatted with ext3, when one has a XP box+VMplayer).

99% of segfaults on my system come from two sources - Adobe flash and proprietary 3D drivers. I finally got so mad at Flash that I removed it. Now I start a VM and run Windows if I really, really have to look at something using Flash. Having unwanted, animated advertising crash your browser is just not a good user experience.

I'm running the Open Source ATI drivers now. They still have a lot of drawing bugs but at least they aren't segfaulting all of the time.

While I certainly had problems with 32-bit Flash a year or two ago I found the "Square" previews to be rock solid stable on my machine. Perhaps the problem you were experiencing was specific to Flash with the ATI drivers. I do use proprietary nVidia drivers on my 64-bit desktop since nouveau doesn't work with my chipset and it has been problem free.

Like many other issues in Linux, different hardware often yields very different results.

Adobe Reader on 64bit is dodgy as well, and as usual no comment from Adobe, which is unfortunately not unusual. It's similarly nasty on Windows. Patching/Updating Acrobat Pro in a managed desktop environment is a nightmare. They still think enterprise users install and update their own software. To put a positive spin on it, Adobe software has turned to crap regardless of platform.

According to this bug, last updated on May 25, the 64bit support is pushed to 11.0

http://bugs.adobe.com/jira/browse/FP-6526

This is from a comment from the bug closer. So no abandonment, just very slow development.

If you reread the article you will find I qualify the word abandoned with "at least temporarily" and "for the time being". Leaving a user community with no updates and known security vulnerabilities does constitute abandonment. It makes it inadvisable to use the product.

We've had multiple promises of 64-bit support before and all have been broken. I'll believe it when I see it.

Flash is just awful, unfortunately HTML5 is still young and we can't yet get rid of Flash once and for all. Even on 32-bit platforms, Flash for Linux is flaky and frequently the Flash binaries crash (which causes Flash elements in Firefox to stop working). Although 10.1 seems to be better. The same goes for Adobe Reader which has turned into major bloatware as of 6.0 and higher. And let's not forget about the bi-weekly security updates for Flash. Maybe Adobe should focus on fixing security before they introduce more features and invite more security holes.

Dumb(?) question. Why run 64-bit Linux at all? Yeah, there may be some limits in accessing more than 4GB of memory, but beyond that, does 64-bit code run any better than 32-bit code? Or does it just take more memory to do the same thing?

Another dumb(?) question. Once Adobe has all the 64-bit conditional stuff it needs in their code base, isn't it just a recompile to create a 64-bit version of each release? Is there something about 64-bit Linux (or X11) that makes it harder to use a single code base than on other platforms? If not, what's up, Adobe?

Whether or not running a 64-bit OS improved performance depends on the application(s) running. There are applications which do take advantage of a 64-bit CPU. If you do real time 3D rendering the performance difference can be huge. Gamers tell me that some games have significant performance penalties when run on a 32-bit OS. Some applications involving a great deal of computation will perform much better on a 64-bit OS. One place I supported used a specialized computation chemistry application that really and truly required a 64-bit OS.

If all you are doing is e-mail, word processing and casual web browsing then you are really unlikely to see any difference at all.

Regarding compiling for 64-bit, since nobody except Adobe can see their source code only Adobe can give you an answer.

Thank you for the article. Hopefully it will remind folks to keep on Adobe's back about 64-bit Flash.

It is disgraceful that Adobe can treat Linux and Free/Open Source Software (FOSS) like crap, and still be accepted as "silver Members" of the Linux Foundation.

Just because most of the Foundation members are focused on business/server environment is no reason for the organization to accommodate the inept, lazy, parasitic behavior of Adobe towards Linux and FOSS. The company needs to be given an ultimatum in regard significantly improving their support of Flash on Linux, or else be kicked out. Nothing less.

Ubuntu 10.10 Namoroka 3.6.18Pre flash player 10.3.181.22 works

32-bit maybe, but there is no 64-bit Flash Player 10.3.x. Never has been, never will be.

News Topics

Recommended for You

Got a Question?