More 'Inter-discipline' questions

Question

Proposal: Embedded Systems Programming and Design

There is already a lot of questions from everyone who doesn't understand embedded to question why our questions wouldn't be on topic on SO and/or EE. So, let's show them by proposing better example questions that wouldn't be a good fit on either site would be best answered on a cross-discipline site. This way, an OP could get answers from multiple angles w/ regards to specialties and areas of expertise. Yes, once the site launches there will undeniably be questions that are on topic on those sites, but it won't help us in the beginning to be on one extreme or the other.

SO======================EE

          ^^^^^^  <<<<<< EMBEDDED LIVES HERE!!

Questions that can't help but involve hardware and software, such as

• assembly
• bootloaders
• kernel drivers
• embedded C that uses pointers
• USB/similar implementation specifics
• FPGA Programming!!
• OS Porting
• Exposing Processor specific features to a kernel

Answer 1

I just had a look at the top-voted questions on the soon-to-be-defunct Embedded Systems site, to find migration candidates, and I do find that between Stack Overflow, Electrical Engineering and Unix & Linux, a majority of questions are covered. The two gaps that I see are:

• Integration questions — the kind of stuff that usually fits on U&L if you're running Linux, but lacks a home if you're running another OS.
• General design questions — they're too “whiteboardy” for Stack Overflow, too concrete for Programmers, too softwary for EE.

In addition, there's the problem of our professional activity being spread all over the place.

While Embedded Systems should exist to fill a gap, it should not be gap-shaped. It is to be expected that a majority of questions will be at least nominally on-topic on other sites — excluding topics that are covered elsewhere would not solve the spread all over the place issue.

Here are a couple questions from the private beta that made my high-scoring-and-answered list and didn't seem to have another home (in addition to some candidates for migration which can survive but are a poor fit).

• http://embedded.stackexchange.com/questions/149/integrating-encryption-security-into-zigbee
• http://embedded.stackexchange.com/questions/162/how-do-you-design-an-embedded-linux-system-for-reliable-field-upgrades-assuming

Answer 2

You've correctly said that the best thing is proposing better example questions that wouldn't be a good fit on either site. As a regular EE.SE user there are probably a few things I agree with on your list, but some that I don't. Running through them and using the newest questions at the moment under each tag so I'm not being biased:

assembly - Doesn't really have to be related to hardware, Stack Overflow has plenty related to user-mode programming for a PC and Could an ARM (ARM7TDMI) Branch instruction take 6 cycles? seems to have fared pretty well on EE.SE.

bootloaders - I've answered a few myself, never seen a close vote on one (apart from generally poor questions) and the latest one How to update MCU bootloader? is answered and has no close / down votes despite being a little vague.

kernel drivers - Agree while not specifically off-topic on EE.SE for hardware drivers it's probably not the best place to get an answer, and harder to search because the tag related to voltage / current drivers. Having said that not all kernel drivers are hardware related (file systems for example) so there might still be a lot of expertise on Stack Overflow that's relevant, but otherwise probably a good one for Embedded.SE when it relates to hardware drivers.

embedded C that uses pointers - Embedded C questions are on-topic at EE.SE as long as they relate either to hardware or a resource limited processor, otherwise Stack Overflow is better. The very last C tagged question was closed because it was a poorly defined request for a UI library that said the hardware hadn't been picked yet, but the next one What happens when you fall off the end of main() in a PIC? is answered / upvoted even though it turned out to be a duplicate. But regarding pointers in particular as long as they have an embedded aspect because of architecture / FLASH etc they normally go well on EE.SE.

USB/similar implementation specifics - The last question What exactly does OTG_FS_SOF do? Is it vital or imperative to USB host mode? doesn't have an answer yet, but nobody is rushing to close it. Some fairly high-level questions like What is a product ID in USB and do I need to buy it for my project? have proved popular.

FPGA Programming!! - Looking for the last one related to programming and ignoring one that was part of a long string of horribly formatted gimme-the-codez questions from the same user the next one is Using Digital Clock Manager with Verilog to generate 25Mhz clock from 32Mhz internal clock. Once again I've never seen a close vote on one apart from the questions that would be unwelcome on any SE site.

OS Porting - Agree this might be a good area as long as you avoid anything too broad. Specific hardware / programming questions though would be OK on EE.SE / SO though. But questions about builroot for an ARM and cross-compiling and such things would be good topics.

Exposing Processor specific features to a kernel - I'm not 100% sure what this means because the kernel will normally have access to specific processor features, but assuming you mean using those features to enhance / change operation of the kernel that's probably another good one that doesn't fit existing sites.

I'm not being negative on the proposal just highlighting you will have a lot of overlap problems to overcome initially. I agree with Gilles that the best questions may be ones that involve integration between platforms and higher-level design questions that wouldn't really sit well on either SO or EE.SE.

For things like Linux hardware device driver programming while there may not be a lot of expertise in the area on EE.SE at the moment it's incorrect to think that experts will magically appear for a new site and I don't think good questions in that area would be closed EE.SE, even if they went unanswered. I think most mature SE sites have on-topic and heavily upvoted questions without answers just because nobody happens to know the answer.

Answer 3

Personally, I think a succesful relaunch will involve showing the experts in the respective groups that there is an advantage to handling the majority of embedded questions in one central site with a concordance of users, and that participating experts will get more out of interactions developed at such a site. Trying to walk the fence, i.e., staking a claim on those questions that don't fit well on other sites, is a strategy that won't bring us far. Yes, there certainly will be questions of that type, but the lion's share won't be.

To put it in @BigHomie terms

Software +++++ + + + + + + + ++++++++ Hardware

SO cov.   ^^^^ ^  ^

EE cov.                               ^  ^ ^^^^

emb cov   ^ ^^^^^^^^^^^^^^^^^^^ ^