Year of the Beagle | Hacker News Reader

Year of the Beagle

beagleboard.org

69 pointsChuckMcM48 comments

Off-topic browser usability rant:

Web developers & designers: Please STOP messing around with how my browser works. In this case, I want to open a link in a new tab, so I Cmd-Click[0] only to have my request ignored with the new page in the current tab ... but then I do get a new tab with the original article loaded into it. This might be fine for your experimentation, but it's not fine to force your incompatible UX onto my browser.

Other things that I've seen lately that are detestable:

- prevent me from zooming on mobile

- similarly, covering your actual content with an ever-present menu or heading (and I can't zoom to get around it)

- change the operation of the "back" keyboard shortcut to navigate articles on your site rather than the actual history in my browser

There are other things that are annoying, too, but they all boil down to some web site developer deciding to repurpose my browser's UI to their site. Please stop.

[0] Google Chrome 39.0.2171.95 (64-bit) on OS X 10.10.1

That's what's going on? I thought it was just a bug in Firefox Nightly or something.

BTW a normal click also does the thing of opening the new link in the current tab and spawning a new tab with the original page.

It's for Google Analytics tracking of links. The links in the page use Javascript:

  <a onclick="goTracked(&quot;link-external&quot;, &quot;http://octopart.com/bb-bblk-000-circuitco+electronics-26478322&quot;);" class="external" href="#http://octopart.com/bb-bblk-000-circuitco+electronics-26478322" target="_blank">BeagleBone Black is readily available all over the world</a>

then the Javascript function in question:

  function goTracked(action, link) {
    ga('send', 'event', 'exit', action, link);
    setTimeout('document.location = "' + link + '"', 100);
  }

So, if you try to Cmd-Click a link to open it in a new tab (or middle-click, which is my preferred method), the target of the link is opened in a new tab. But the link's "href" target is just a fragment, so it opens the article in the new tab. 100 ms later, the current tab is redirected to the "actual" link target.

I absolutely love my beaglebone, aside from some issues with the Linux drivers for the USB host port, it's been very reliable and useful.

To date, I've used my beaglebone to

- stream live video of my baby kittens from the window well they were born in outdoors (their mom was a feral cat, I used the video stream to keep track of them until they could be rescued and brought inside)

- transparently encrypt USB drives using the client+host ports [1]

- use the onboard hardware RNG and some custom python[2] to supply entropy to other boxes on the local network

- use it as an offline environment to perform sensitive PGP and CA signing operations

And I'm toying with the idea of building some kind of open source network/USB connected hardware security module system with the BeagleBone as the reference platform to run it on.

[1] https://news.ycombinator.com/item?id=8713155

[2] https://github.com/infincia/netrng

"... I absolutely love my beaglebone, aside from some issues with the Linux drivers for the USB host port ..."

How did you go installing Linux? What distro? Have you got wifi working?

I grab the latest ubunutu of the elinux.org site, and Wifi works out of the box (with a generic Wifi USB device). You need to edit /etc/network/interfaces and make sure that you auto wlan0 and provide WPA passwords and what not.

thx @ChuckMcM I've only read about unstable installs - the only reason I want to try a beagle is the OpenBSD armv7 port http://www.openbsd.org/armv7.html can't see wifi though :(

Do your research on what Wifi devices to use. Many don't work and others are unstable causing agony and frustration like a thousand burning suns. After all you rely on drivers that compile perfectly on ARM.

This one works well: TP-LINK TL-WN722N

I've been hugely impressed with what the BeagleBone offers; it's really a shame it frequently gets overshadowed by "that other board from the UK". Apart from just having a vastly superior array of I/O options (including analog in, which is really important for sensor arrays) the fact that the design is fully OSS is really encouraging. I hope the community around it continues to grow.

However, the GPU on the beaglebone is much more closed and undocumented than the one on the Broadcom chip.

It's really expensive though :( If they offered a slower version at a price comparable to "that other board from the UK" (starting to sound like Voldemort) then the community base might grow similarly

I agree, I only have a Pi because it was free, I have a Beaglebone because I badly wanted one before they were even released for sale :D

Aside from all the I/O pin options, that USB client port is particularly valuable paired with the Linux USB Gadget drivers.

It would be even more successful if it was possible to actually buy the boards.

I got one early on, and I would have more, but every time I've looked since they've been out of stock. A big part of the RPi success story is that they figured out early on that supply chain would be a problem and got the right partners involved to deal with that.

So are you outside the US? I have bought all of mine from Digikey, they currently have 787 in stock (http://www.digikey.com/product-detail/en/BB-BBLK-000/BB-BBLK...) so presumably if I ordered today I'd have it in 3 days (priority mail)(Mouser has over 1200, Element 14/Farnell has eight thousand)

[1] http://www.mouser.com/ProductDetail/CircuitCo/BB-BBLK-000/?q... [2] http://www.element14.com/community/community/designcenter/si...

They were definitely out of stock this summer, when I was looking at updating my BB project to BBB.

Availability is much improved over just 6 months ago.

My beef is now that I've actually got one in hand, it's like pulling teeth to find documentation on addressing/using the GPIO ports. I can easily use them with the Node libraries, but where's the docs on setup/use of GPIO with the C runtime if I don't want to use Node?

Working under pressure to build a prototype for a device, I ran into the BeagleBone around Dec. 2012. I was really impressed with the fact that they open sourced all details related to the board. TI has provided extensive documentation for the AM335x processors.

Sweet spot - I think TI hit the tinkerer sweet spot with the BeagleBone, which was smaller and less powerful than its predecessor, the Beagle Board. The open nature of the board itself adds considerable value as it can be used as a reference design.

Ease of modification - For my project, we just replaced the HDMI and audio connectors with an extra Ethernet and a few micro-USB connectors and rolled out our own custom board... all in the span of a few months.

Ease of use - They need some work here for the average tinkerer who might not be familiar with the Armstrong Linux distribution. However, that just worked perfectly for me. For anyone interested, the Open Embedded Core or the Yocto project would be better resources to look at. Debian and Arch have Beaglebone focused distributions too. I would recommend against developing code on the Beaglebone itself.

So for someone looking to create a customized "device" of any kind, this would be a great choice. Also, I sell myself out by the hour doing this kind of stuff :). I would be happy to help/consult with anyone if interested.

USB is still problematic on Linux. In particular the musb is completely unstable when "babble" is detected. You have to use the right kind of USB hub (or modify the cable) to keep power from backfeeding into the BBB. It seems your experience is different from mine.

We've had the same experience on BBBs re: powering USB devices. Using a powered dlink hub helps to resolve it (of the 10+ brands we've tried).

Initially we did have some problems. Most of it was due to not being able to source enough current. We haven't seen any issues so far. Thanks for reminding me. I'll keep an eye out for that.

I am confused by this article. I had always been under the impression that Beaglebones were not approved for use in commercial products. I'm almost positive that I've read that in their documentation. But here is the creator of the project happily pointing to companies using what I assume are off the shelf boards in their products.

What gives?

What sort of approval would you need to use COTS in a commercial product?

Do you mean this:

We do not encourage the use of the board that we manufacture under the BeagleBoard.org logo in commercial products. We are not able to schedule parts and arrange for production for orders that we cannot see. Meeting demand is difficult as a result.

I've got several Beaglebone Black boards and find them a wonderful way to drop a computer into something. My next challenge is to build a custom board with their design.

The way outbound links are handled on that page is really strange....

Thank you, Beagle!

And a big thank you to TI, for this project. I hope that if TI views their involvement in BeagleX as successful, they can extend what went right with this project into other product lines (open design files, reasonable pricing, accessible documentation, accessible IDEs/compilers, transparency with design decisions, and a true sense of community). My company has begun transitioning to Nordic for RF and it feels like dealing with a startup (Nordic) vs. an incumbent (TI).

The ability to run android on BBB is a wonderful advantage of the BBB vs RasPi. The Beaglebone was my introduction to exploring embedded android and it was perfect for it. The community is smaller than the RasPi but it is very strong. I hope there are more tools supporting the BBB as a point of entry into android.

why couldn't the guy just mention Raspberry Pi in the article? Why did he have to refer to it as "that other board". Comments like this disturb me when I hear them coming from the hacker community, which is supposed to have a foundation of openness and selfless sacrifice. If something is considered a valuable asset to the community, give props where props are due. Don't say "that other board" just because you don't want to give Raspberry Pi any more market share. Market share and competition should have no place for discussion in the open source community.

I don't think he was seriously trying to conceal the competition -- everybody knows what he means (and has already bought an RPi if they were interested). I read it as a teasing form of friendly rivalry. And you can't go create a low-buck board designed to sell in the 6 or 7 figures without inevitably comparing yourself to the RPi. When it comes to getting hackery Linux computers into people's homes, the RPi is the gold standard, and comparisons are inevitable whether you're a competitive person or not.

"that board isn't open hardware: there aren't open source layout files or open source editable schematics and the components in the bill of materials aren't readily available"

This is talking about Arduino of course, and I do wish they would stop using the word "open-source" if they don't bother to document the hardware they "open."

Throwing a half-documented Eagle file over the wall doesn't count.

If they really want to open-source the design of the arduino, they'd publish an official Bill-of-Materials for the project.

Edit: They'd also include official gerber files

They're referring to the Raspberry Pi. The Broadcom chip on it is not open or documented.

The Arduino is from Italy.

Ok, you're right. I've never had a problem with the Pi because they've never claimed to be "open source"

The GPU on the other hand (which is part of the chip) is much more documented on the Raspberry Pi than the Beagle Bone.

They're talking about the Raspberry Pi not Arduino. Raspberry Pi has always said publicly that the whole thing was never likely to be open source.

You're correct. I jumped the gun and pointed to the Arduino - which is claiming to be open source. I've never had an issue with the Pi because they've never claimed to be open source.

I have a crazy theory: There is no such thing as open source hardware. It's a myth. If a project has any electronics, there's a closed bit in there somewhere. Even if there are published layout files, the chip is still closed. Can anyone fab their own chip? Same goes for BeagleBone, which drives demand for (closed) TI-based ARM chips.

This is true, but given the Arduino does not publish Official gerber and BOM, there's zero chance to recreate the hardware without guessing/reverse engineering. The specs are not version controlled for public consumption.

So the arduino "platform" is not open in that you/I do not have an offical specification to work from.

What about the BOM can you not derive from buying an Arduino and looking at it alongside the Atmel datasheets?

Sure, it might be easier and $25 cheaper for you if they published a text file, but it's not like you have to reverse engineer a spaceship to make an Arduino clone or derivative product.

Evidence: there's a wide selection of Arduino compatible boards on the market today.

IMO, publishing the Eagle file is better than publishing the Gerbers. It's roughly the equivalent of publishing source code instead of binaries.

I'm genuinely curious what isn't open-source about Arduino in your opinion?

It's great they provide the eagle files. However when you generate gerber files there are often a lot of settings to configure depending on how you want the files generated. So it's certainly possible for someone (especially a noob) to generate the wrong gerber data set for a given eagle file.

It would be great if they included the eagle file, official bom, and offical gerber data set all in github.

I suspect they do this internally, but they certainly don't give the public access to this information.

I really can't figure out what would satisfy you regarding Arduino. I found a full schematic (pdf) in a few seconds of searching. It has all part numbers labeled. http://arduino.cc/en/uploads/Main/Arduino_Uno_Rev3-schematic...

A simple google search for "arduino bill of materials" found several hits from text, to json, to Excel formats.

If the schematic was only available as pdf, I could understand the complaint that it is not an editable format. There's also Eagle files of schematic and board layout. From that one can make gerbers in a few minutes. Or modify and make something completely different. But in a different comment you said that Eagle files don't count? Why is it that the specific thing you request doesn't count?

The eagle schematic doesn't not have all the mfg part numbers listed. Lots have to be reverse engineered - a good example is the pushbutton switch.

Edit: Also, if you read the first page of their schematic they pretty much say don't use this schematic as a legit source. So if it's open source, my question is where is the "real" source. Answer: they don't publish it. It's locked up internally in the company under a private repo.

On the schematic on the arduino.cc website, the pushbutton switch is labeled "TS42031-160R-TR-7260". A quick google search leads to Omron as the switch manufacturer. Even so, picking a simple button switch from digikey with the right footprint is a trivial task.

http://arduino.cc/en/uploads/Main/Arduino_Uno_Rev3-schematic...

The first page of the schematic has a standard disclaimer. It's not different than the standard disclaimer at the top of every open source software source file. (Hardware version: "Reference designs are provided AS IS and WITH ALL FAULTS ....", Software version: "THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND ...")

In the world of electronics, it is normal to label a resistor simply as "10k". Is the resistor from Panasonic? Vishay? Yageo? Is it thin film? carbon? ceramic? It doesn't matter and I don't want to see all that info on a schematic.

The real source is right there on the website. I've personally used their reference designs to make arduino-compatible boards, so I'm positive that all the information is available.

I'm still at a loss as to why you believe that Arduino is not real open source? Is it just because they didn't put a BOM on the website alongside the schematic?

A person can't print out 3d models without a 3d printer. The models could still be said to be open source.

Or taken even further, a person without access to a hammer would have a very difficult time following blueprints for a house.

Open source is really just "given tools and knowledge not specific to this particular case, you can replicate everything we've done". For any sufficiently niche application, "tools and knowledge not specific to this particular case" starts to become an increasingly large hurdle, but I'm not sure I'd go as far as to call open source a myth with regards to hardware.

This is why software is great. If someone screws you on the hardware, you port to another platform.

>Throwing a half-documented Eagle file over the wall doesn't count.

The eagle sch/brd files are the source.

>If they really want to open-source the design of the arduino, they'd publish an official Bill-of-Materials for the project.

You can export the bom from the sch file in eagle with the bom.ulp script.

>They'd also include official gerber files

And you can generate the gerbers from the brd files...

These arguments are ridiculous and are akin to saying, "this software project isn't open source because all they provide is a the source code and build scripts!"

With all due respect, have you tried generating a BOM out of the "source" files? You can't because a huge portion of the parts don't contain the mfg. part numbers.

If they had provided all this info in the eagle files, then yes - it would be a valid file.

They conveniently left it out. I guarantee somewhere in the company they have a complete BOM for the arduino, but they certainly don't publish it on their website.

The only BOMs I've found have been through reverse engineer community efforts.

The arduino hardware is so trivially simple I can't see how anyone would see this as a problem. What you're complaining about is a non-issue.

It's not. Take for example the pushbutton switch on the arduino uno - it has a very specific one-of-a-kind footprint. So if you don't have the mfg part number, there's no way to populate the PCB. You're literally going to have to re-layout the board and choose your own. And what if the debouncing of the new switch is different than the original and doesn't work with their firmware.

Providing the mfg for ALL the part numbers is critical in hardware design. In a lot of cases, where hardware is mission critical (like space or medial), if one single mfg part number changes or the source manufacuture changes, the entire system needs to be requalified/tested because of the larger ramifications.

If you start willy-nilly grabbing parts off the shelf that might work, you're going to eventually get burned.

My request is simple - if you're publishing an open source projet, please publish the offical BOM that the platform was validated under, and put the BOM in source control should a part become discontinued and need a replacement. This is standard for any hardware product.

Edit: The switch mfg part number is actually a real problem. See this person who tried to reverse engineer the part number. Notice the ??? in the mfg? Google doesn't return much with that part number. If you can find the actual mfg datasheet for that part number I'd love to know. I have no idea where to order that switch.

http://mburdis.com/BOM_Ardiuno_UNO_Rev3.html

Beagle owner here, I owned quite a few revisions.

You can stack up to 4 caps to make it a product quickly.

However I'd prefer a SODIMM module for volume production instead of the connectors used on BBB.

I once chose the BeagleBoard for a startup's product prototype. It had just the right hardware features, ran Linux, part price ok (esp since only needed 1-3 at most, on near term horizon), plus the "not tested/blessed for commercial public embedded projects" was not a showstopper for a demo prototype. If I had a similar need today I'd prob choose the RaspPi, due to combo of lower price and bigger community. But RaspPi had not been out back then.

Reply on news.ycombinator.com