64 comments

[ 4.1 ms ] story [ 122 ms ] thread
seems too expensive for the spec!
It's battery powered, they have to make a compromise somewhere.
I want to see our devices to work with input power when available and use battery only when there is no input power.

In other words, a device must be able to power on and function correctly without a functioning battery plugged in. I think all notebook computers pass this test. Why can't phones and tablets do the same?

I got the beta version and it died already. Doesn't charge and doesn't boot.
How did they handle the warranty?
I wrote a few days ago to the support. Haven't heard back from them, yet.
Very sorry to hear that, I'm sure someone will be in touch shortly. Are any lights showing or is it completely dead?
I hope so.

There was a green LED and a second one blinking during charging. Eventually the LEDs died and now it's dead. Strangely it booted after many attempts once, and then it gave up ever since. Power cycling has the LEDs all blinking twice, then they are dark again. I tried different chargers - makes no difference.

same with my unit. It does not boot anymore, or only after trying 10 minutes pressing the button to get it to boot.
In their FAQ, comparing an Ocean to RasPi, they say the RasPi has no WiFi or Bluetooth - which was true before Pi3, but is not true any more.
The BattPi Kickstarter didn't succeed last year. This is better how? Also, the PINE64 has a built in battery charger. The ODROID-C0 has a battery option too.
This looks like an overpriced Raspberry Pi with wireless charging.
Why not use a pi and a powerbank?
Getting a nice case that encloses everything might already be a very valid argument for everyone who doesn't want to tinker with that. Integrated storage would also be a plus for me, because my PI corrupted lots of SD cards.

The other typical competitors would be NUC style products, but the differentiation point here is the included battery. So it seems to have something unique.

That's a huge point. I've said a few times there's a gap in the market for open hardware with consumer-grade finish.

My only Arduino-based home automation component sticks out like a sore thumb. But all my proprietary light switches fit beautifully into my walls.

However, the internal storage in this device is "8GB-64GB Internal Micro-SD", so you'll probably have the same problem with corruption.
Yeah, I discovered that also, and even changed my response. Then I discovered that it has 4GB Flash plus that SD card - so I changed it back, with the expectation that the integrated flash is somewhat more reliable. But of course - it's only 4GB.
This is Packaged up neat enough in a manner that ensures fewer awkward questions at airport security.
One of the questions in the FAQ (and repeated in this thread) is why not just buy a RasPi - In the '80s and '90s I was working for companies that manufactured electronics and, unless you're producing very large numbers of units, the case and the power system were always our biggest cost.
"mobile" as in Wi-Fi, not 3G/4G.

Seems awfully expensive for its specs. You can get pcDuino3 for $66 with the same CPU/RAM.

EDIT:

Raspberry Pi 2 has Cortex 7 and 1GB of RAM too, and it's like $34.

Banana Pi also, and it's around $31.

Honest although tangentially related question: why node.js?

I honestly really don't get why someone want to use js serverside, and I say that as someone who did create cool things with js client side and wants to know why I should spend time on this new cool stuff.

Edit: automatic upvotes for serious answers as long as I see them before going offline :-)

There are many reasons, but I think the simplest is that using a language in house has a very real and non-trivial cost. Expertise, coding standards, library vetting, etc.

By using Node.js in a Web shop, you can eliminate one language.

Other benefits include re-use of client side code, async everywhere, and popularity.

One of the promises of node.js (that I've not seen pan out really in practice since) that got me interested 4-5 years is the ability to render a web page on server or on browser using same codebase. Take as an example a heavy codebase like Yahoo! Mail or Gmail. Notice how slow Gmail especially is from login to "ready state"? Now imagine if the first page of your inbox was rendered server side and sent as normal HTML markup. If it was designed well, the JS would progressively load after that and if for example I click compose before that JS was loaded yet, the click would be treated as a normal click and it'd round trip to the server and then render the compose view and send me back the new page. If the JS had loaded, it'd just be rendered in browser like normal. Doing tricks like this can massively increase the perceived responsiveness of web apps. I've seen prototypes of Yahoo! Mail built like this and man they are so fast, but sadly doing this is well is very complicated and node.js never actually helped as much as it promised too. Don't get me wrong, what I described above I've seen done in production code, it's just it was a very manual and slow process and node.js never did that much to help, IMHO.
You might be interested in what is going on in Drupal 8 with regard to pre-caching and a newer (now in core I think) module called BigPipe. It does some of what you describe in PHP in a fairly generic way. AFAIK, you just enable this module and ensure caching is enabled. From that point on, slower assets load when ready so the rest of the site is already visible... like Facebook.
I'm more talking about fast start SPA, not partial content loading strategies. There's a number of utilities that does what Drupal is doing. Turbolinks has been doing what you describe for years (if memory serves) with Rails.
Ember FastBoot? And React also published something similar recently.
I've seen many attempts, but nothing has really taken hold as its really difficult to do right. Rather than pointing at libraries that can "help", I ask you to instead point at any places it's being done well in production, especially at scale.
Why JS? Because most developers are JS developers. [1]

To business decision makers, that looks like a better strategic decision - Now my web app devs can do server side! Obviously this is totally wrong, any decent developer can learn a new language/framework. The architectural concepts are totally different on server side, which is what HR should be hiring against. But hey, what do I know.

1. http://stackoverflow.com/research/developer-survey-2015

Most IT-based business hire based on the lowest cost option and what tech stack the founders are comfortable with. Probably not as true in SV and SF, but the rest of the country it is.
Node.js is event oriented I/O. It lets you design some server tasks in a way that is non-blocking and thus more efficient.

(Some other languages let you do that too)

Pretty much all general purpose languages let you do that. Only Node makes you.
PHP doesn't
Yeah OK :) HHVM is not the usual PHP runtime one expects to build the average web service upon.

Anyway, my point was that Node.js main core value was to let people script evented I/O over a lightweight runtime.

When Node came out, it was something new.

Sure. I just happen to agree with the notion that node.js is mostly unique only in that it pushes you down that path.

Java has a fairly rich NIO path that you can choose to take or not, and predates node.js. I believe the Tornado package for Python predates it as well.

Certainly, there's some advantage there...fewer surprises with 3rd party libraries that aren't non-blocking. But, there's downsides as well.

Why did Java take over in 2000?

Because starting in the mid-late 90s, B- and C-rank universities began adopting Java over C++ or Pascal as a teaching language, producing a glut of graduates who knew Java. As IT needs grew because of the internet boom, companies built their systems with the language that the available programmers knew.

These days, more and more "teach kids to code" initiatives are starting with JavaScript because it's easy to get started with, existing in every browser known to man. With IT infrastructure in for another period of growth/change, especially in startups, it makes sense to build out in JavaScript because it's what incoming fresh programmers know.

It doesn't make much sense to me, but that is how they do in the corporate world.

"Raspberry Pis do not have built-in Bluetooth or Wifi"

Time to update the page. http://makezine.com/2016/02/28/meet-the-new-raspberry-pi-3/

Yeah but the range and data throughput of the Pi wifi/bt chip is nothing compared to a "real" USB stick with a proper antenna (or, in case of WiFi, multiple antennas).

I hope that this gadget has a "real" antenna embedded instead of a PCB one.

I manage various RPis and had more than a few mini-devices acting as Linux server (ebox 2250, bifferboard, etc.).

I would give a shot to this one if it had an SSD, dedicated GBit Ethernet and/or GPU unit, in short if it resembled closely a mini-scale real linux server somehow.

When that doesn't happen I'd go with a device that's more open, cheaper and has better support and bigger community, such the the RPi.

I think for these requirements and no need for running on battery the NUC device range should fit really well.
Given the fact the communication infrastructure is super deployed (wi fi everywhere, networks everywhere), why would I want to actually move my server ? I mean, the client moves and the network infrastructure make sure the server is reachable anytime anywhere... So why would I want to move it ?
There are still plenty of places with no WiFi is available, i.e. marine and farming environments (the areas I work in). So I could see this being used as an IoT hub. But I'd rather just use a Rpi.
I was involved in the development of this with two stellar engineers (David/Kousha) at iCracked. Surprised to see this here at this point as all available units sold several months ago. Happy to answer questions!
Apparently this device seems to be based on Allwinner A20 SoC which nowadays has quite good mainline Linux kernel support.

So I wonder why they are still using vendor-provided 3.x kernel.

Having built my own portable server a couple of times* using old Android phones, I genuinely like the idea, but find it quite surprising that someone actually went out and built an integrated device - must be a pretty small niche market.

*: https://taoofmac.com/space/blog/2013/04/28/2330

I do think there's a market for a RPi type board with a high quality case and an integrated battery. This seems to fall short though, in a few places...

- Marketed as a portable headless server, and indeed, limited to something like that. No access to GPIO, no video. But, relatively low 1GB ram, and limited, non-upgradable storage.

- Wireless charging seems like a feature nobody is looking for in this type of device. Charge time with the wireless is 10 hours vs 4-5 hours via USB. I would guess it's also driving part of the price point.

To me, if you're going to market as a server, you base it on something like the CubieTruck, more RAM, 1G wired ethernet, more storage options, SATA, etc.

Or, go the other route and provide what's good about an Rpi device (GPIO, video, low cost) and add the quality case and battery.

> non-upgradable storage.

It is upgradable. It says on the site you can use a Tork to open it and switch the SD card.

There is still plenty of room for quality linux laptop for devs - this is much bigger market than this. Question arise is it really that hard to do ?
Not sure what is the purpose of this. No ethernet, only 1GB of ram and only 2x 1Ghz allwinner CPU for $150... RPi/odroid has bigger community and is more powerfull. And if you need something in nice looking case, there is a lot of chinese atom based mini PC with buil in battery for around the same price...
>> there is a lot of chinese atom based mini PC with buil in battery

Had me curious, so I found 3 different types of these. Interesting that they all come with bundled Windows, but are still pretty inexpensive.

http://www.aliexpress.com/item/Vensmile-IPC002-MINI-PC-mini-...

http://www.aliexpress.com/item/Windows-8-1-smart-tv-box-mini...

http://www.aliexpress.com/item/Windows-10-Mini-PC-with-5000m...

Look up VIA's mini PC's too. VIA has been shipping low-power, multi-core, VT-enabled, crypto-accelerated x86 for a long time in boxes that were $100-300 new. You can always find some on eBay, too. I used to use them for physical decomposition of secure systems with non-DMA links & KVM switches.
I would love this if airlines allowed personal wifi in-flight so I could offload my development builds which I currently have to run on my laptop. As it stands, you can either use the airline's onboard wifi or just not use wifi. :(
Wouldn't any somewhat recent laptop be much more powerful than this anyway?
I think a Pine A64 with a Lithium Battery is a better solution(you can a battery bank while using a Lithium Battery, which also allow hot swap of the battery bank)

Since you can get WiFi, Bluetooth and Gigabit Ethernet.

After which you just need a pretty case and you are pretty much set.

Ads are not desired on hacker news. At least put some content around your advertisment that enables people to learn something new.

And what's the difference between this and all the other small linux machines that you can buy, like raspberry pies? From the landing page it looks like the developers don't even know that these kind of computers exist since more than 5 years now.

Rasp is explicitly mentioned in the OP.

Why don't I just use a Raspberry Pi?

The Raspberry Pi is a fantastic product, but it is built with an emphasis on education and 'home hacking'. It lacks certain features found in the Ocean. Raspberry Pis do not have built-in Bluetooth or Wifi, and do not come with an on-board power source. While you can purchase cases for them, these cases are rarely designed with portability in mind. Exposed GPIO pins tend to bend in your pocket. A Raspi can be augmented with additional hardware to equal the feature set of the Ocean, but this hardware will compromise the portability of the Pi.

Hard to misconstrue this as an ad when it's not even possible to purchase the product right now.
Then I'm happy to tell you that all three slots on the linked page (which is a sales page) are sold out now. Last time I checked it was only the first slot.
The only use case I get from reading the website is bring a server to a remote location with no wallpower/connectivity. I'd argue that is something that nowadays is solved with software, not hardware. We've done data interactive offline first apps that sync when connectivity is available. This approach means that client and server run on the same $50 smartphone with the same CPU/RAM/HD as these severs but with many more features.
For $15 you can get Orange Pi PC + add $x for battery and WiFi and you are much better off.
This is one more ideal use case for an always-on VPN, so you can always reach the server privately, securely and potentially on the same IP address.

Are there any plans to integrate something like (Disclaimer: I made this) https://wormhole.network with it? Actually, you would just need to include SoftEther's VPN client (https://www.softether.org) by default :-)

Key question: Why would I choose this over setting up a chroot on an Android device?

My rationale is that Android devices are far cheaper, especially older phones second hand. Something like a Nexus 5 can run Kali Nethunter which gives me a full Kali environment if I want it, or I could run a Debian chroot using one of the many various options available.