All features at once! Make EyeFi open source!

[rmore]

In reading through the myriad of requests and innovative ideas that people have for using the Eye-Fi, I was struck that the maybe the solution to ALL of these requests is to make the Eye-Fi system open source! If Eye-Fi published the source for both the firmware and the manager then the open source community would jump on it like crazy. You would see an explosion of features and niche capabilities developed (IMHO).

Perhaps I don't know the Eye-Fi business model, but is seems to me that the bulk of their revenue is from actual hardware sales. Making it open source would, I believe, cause a huge increase in sales as people found more and better ways to use the technology.

Probably I am missing something, and please correct me if I am being naive, but this seems like an obvious way to expand sales and uses for the Eye-Fi hardware with little or no effort on the part of the Eye-Fi development staff.

[Francois]

Going open source, or releasing an SDK of some sort would be something good.

As of now, if for some reason the company goes under (because all manufacturers decide to put Wi-Fi features in their cameras), our Eye-Fi cards are just as good as the plastic they were made from... I know this is not something the administration will want to consider, but it is fact. The manager and the cards rely heavily on a connection to their servers. This is an additional cost to them which doesn't bring any money in.

Last time I went to my local Staples, I noticed a host of cheap digital picture frames and printers with built-in Wi-Fi. It's only a matter of time before it comes to digicams at no additional cost.

If I could get my hands on a simple to use DLL or OCX, I could use it with a bit of Visual Basic (I know it's a "bad" language, but it's pretty much the only one I know how to use rapidly and efficiently) to whip up something... And I know others would be able to do even better in a more optimized language. But as of now, our hands are tied.

For what it's worth, I noticed the manager uses Visual C++.

[MarkW]

There are some open source alternatives to Eye-Fi Manager - see this forum thread.

[rmore]

There are some open source alternatives to Eye-Fi Manager - see this forum thread.

Thank you. I had seen that too. I think there is a big difference between reverse engineering some limited functionality (and having the constant danger that the vendor will make a change that stops your code from working) and having the Eye-Fi actually endorse and support open source and third party developers.

I believe many-many more developers would embrace the Eye-Fi if there was a formal developers program.

[Francois]

These are still just some kind of hack...
Besides, one of the authors described that his CGI script stopped working since the recent firmware updates (back in November 2008).

What we need is a decent solution which is applicable to all platforms. There are good open source cross compilers out now so there's no real reason for some "usable" and working code to be released. The only thing I could see is that part of the code is licensed to some third party and under copyright law. If that's the case, I think the company should be clear about that so people can direct their attention to what can be done to improve the system and features provided.

Eye-Fi could still release firmware updates in a closed system using a dedicated updater and separate the manager part which could be user developed.

[MarkW]

The server side is the side that concerns me the most after going through iMate's MomentoLive shutdown; so that's why I'm pleased to see the excellent server progress.

As for the SD-card itself, the comments from IkonTools' teardown say that the Atheros chip probably runs the already open-source eCos. If the Eye-Fi actually does use eCos, then source for the core card OS should be available on request under eCos' modified-GPL licensing. Other programs that might be running on the card, using the OS, may have other licenses. Work on eCos with EyeFi has been reported over at the EyeFi Linux Hacking blog.

An SDK and assistance from Eye-Fi would certainly help get things moving more quickly though.

Oh, and there are more teardown pictures on Flickr: http://www.flickr.com/photos/sniperninj ... sassembly/

[Francois]

I feel releasing full code might be a first step do do before people simply reverse-engineer it. Hacking sometimes leads to unpredictable results which could cause simply more damage on both the cards, manager and the eye-fi servers. Releasing proper procedures would save a lot of trouble.

[rmore]

I agree completely. Having Eye-Fi support the developer community through a developers program will lead to an order of magnitude more developers working to create interesting applications!

[MarkW]

I emailed info@eye.fi requesting a copy of the source code under the source code licensing agreements, but have not heard back yet. Have other owners requested source code? If so, did Eye-Fi respond?

[berend]

I emailed info@eye.fi requesting a copy of the source code under the source code licensing agreements, but have not heard back yet. Have other owners requested source code? If so, did Eye-Fi respond?

Mark,

Your request to info@eye.fi has not been forwarded to me, but you are not the first one to make such a request and I will provide the same information I've provided in the past. The Eye-Fi firmware is composed of a ROM-resident proprietary BSP (board support package) for the AR6001 chip, the eCos embedded operating system, a proprietary WLAN driver from Atheros and Eye-Fi's own application code. The eCos license stipulates that:

"The license does not require users to release the source code of any applications that are developed with eCos. However, if anybody makes any changes to code covered by the eCos license, or writes new files derived in any way from eCos code, then we believe that the entire user community should have the opportunity to benefit from this."

We take care in making sure that the Atheros driver and the eCos are not integrated in a manner that makes it a part of eCos (in other words, it does not appear as a device driver that abides by eCos device driver interfaces in the linkage). The Eye-Fi application itself uses eCos at the well-defined interfaces exported by eCos and remains a stand-alone application per the above definition. Where we've made improvements and bug fixes, we have contributed back to eCos (and lwIP, the TCP/IP stack included in eCos):

http://savannah.nongnu.org/bugs/?func=d ... m_id=21582

Beyond the baseline eCos distribution, we've applied the SACK (selective acknowledgment) patches for lwIP TCP/IP found here:

http://edin.no-ip.com/legacy/?q=comment/reply/41

At this time we do not have plans to open-source the firmware for the Eye-Fi Card.

I apologize in our delay in getting back to your inquiry via the info@eye.fi alias.

Thanks,

Berend

[MarkW]

Berend,

Thank you for the quick reply, the details on the origins of the source code, and for a great product! While I understand that the whole set of software might not be open source, I am interested in obtaining copies of the parts that are open source. The eCos and IwIP repositories have many revisions and I would like copies of the exact source code that goes with my Eye-Fi 4GB Anniversary Edition (the Eye-Fi Manager seems to update it fairly regularly too); obtaining the source code directly from Eye-Fi will be the most expedient solution.

Thanks!

Mark

[berend]

Mark,

Thank you for your interest, but as my earlier post indicates, we are not obligated to release sources for any component given our adherence to the licensing terms and at this time we are not planning on releasing it for other reasons. Your initial question was about the Eye-Fi Card, but in your follow-up you mentioned the Eye-Fi Manager software as well. The Eye-Fi Manager software uses various components distributed under licenses that do not encumber us to release our source code, either. Chris Davies was kind enough to point out that the the license for OpenSSL (which we just incorporated in this last release) requires us to attribute that "this product includes software developed by the OpenSSL Project for use in the OpenSSL Toolkit (http://www.openssl.org/)". We missed that and we'll remedy in an upcoming release of the software, as well as on our web site.

Thanks again,

Berend

[MarkW]

Berend,

The quote from your earlier post isn't part of the eCos license but instead is explanatory text. The exception in the eCos license says

if other files instantiate templates or use macros or inline functions from this file, or you compile this file and link it with other works to produce a work based on this file, this file does not by itself cause the resulting work to be covered by the GNU General Public License. However the source code for this file must still be made available in accordance with section (3) of the GNU General Public License v2.

I'm just asking for copies of the eCos source code files per section (3) of the GPLv2. I'm not requesting the "other files" nor "other works" (files that link with eCos files such as the Eye-Fi application or the Atheros driver).

Thanks,
Mark

[berend]

I'm just asking for copies of the eCos source code files per section (3) of the GPLv2. I'm not requesting the "other files" nor "other works" (files that link with eCos files such as the Eye-Fi application or the Atheros driver).

Mark,

Thanks for clarifying. I understand now and my apologies for the confusion. You actually are the first one to be interested in our redistribution of the baseline eCos source, which as you point out is an obligation and we can remedy that. If you can give me a day or so, I'll make it available on our download server.

As an aside, and in my humble opinion, not getting a response from me after 1 AM (I do sleep occasionally) should not have warranted an e-mail to others at Eye-Fi, especially where the subtext may be perceived as antagonistic given PR folks can't be expected to have any technical insight into these matters, but I guess we're all entitled to our own measure of tact.

Berend

[hyachts]

Wow, MarkW. Congratulations on flexing your legal muscle and forcing EyeFi to show you copies of redistributed (unmodified) code. What's to be gained by this - knowing what parts of the underlying code EyeFi is using? Seems like a long way to go to prove a pretty petty point about open source licensing, particularly when it sounds like you've been pretty antagonizing about the whole thing.

Seems like those desiring collaboration with Eye-Fi on this issue would be better served with a less aggressive approach.

[rmore]

I am sorry that my original thread degenerated a bit there. I hate to see antagonism (real or perceived) creep into it. Eye-Fi has always been very responsive to my questions and concerns.

I truly believe it is in everyone's best interest, Eye-Fi most especially, if some sort of developer program or open source repository were established. There is definitely interest.

My company would be willing to pay for membership in a developers program, libraries, APIs, development support, etc. if it meant that we could develop (or contract to have developed) the functionality that would allow us to use the Eye-Fi for our field data collection operations.

[berend]

Mark,

The eCos source distribution that the Eye-Fi Card firmware is based on is available for download at:

http://support.eye.fi/open-source/download/ecos.tar.bz2

Sincerely,

Berend

[MarkW]

Berend,
Thank you for your quick replies and for this set of source code. I hope that community contributions will continue helping to improve Eye-Fi's products!

[electronrancher]

Hi Berend,

I downloaded the source code, but unfortunately I believe there are a couple of things that might have been forgotten.

1) Hal and hardware definitions - It's OK if you truly have a standalone piece of software that can be compiled and flashed seperately to provide cleanroom interaction with eCos, that file does not need to be released. But such an outstanding feature would certainly require some heavy modification to the eCos HAL layer at least. This modification, and the ability to hide all vendor hardware specifics would certainly be seen as a great improvement to eCos, and the modified code would need to be released.

2) Bootloader. Are you absolutely sure that it is in no way related to RedBoot? Specifically, if I went tracing around I would not find any sequences of instructions that appear highly parallel to the boilerplate redboot startup?

Thanks in advance for the clarification!