Silicon Valley Redneck

No website for old engineers

Got this project that's been using a Raspberry Pi Model B... and for the next stage it looks like we want to go to the B+, for reduced power consumption, more USB ports, and that kind of stuff.

I want specs for the critter, so I know what I'm dealing with.

The Raspy Pi Foundation's website is... unhelpful.  Takes some poking around to find the product page (there not being a "Products" tab), and actual specs for it are nowhere to be found.

I found some documentation via Lady Ada and SparkFun.  Not much, but at least somewhat helpful.

But... no hardware spec document.  No schematic.  There's a fragmentary schematic showing approximately what's connected to the various connectors, but it's missing some important details, like the identity of the new switching regulator chip they designed in.  (I'm interested in the maximum input voltage it'll accept, and also whether the +5V rail goes anywhere other than the regulator and the expansion header.  Exceeding 5V would, at minimum, require removing the TVS diode.  Oh: I guess it must go to the USB subsystem, no?  So it's gotta be right around 5V.)

And, getting back to the website: it appears to have been designed by the marketing department, for children.  Or by children, for a marketing department.  One or the other.  Lotsa happy-shiny, and little readily-found substance.

Oh, well.  I pretty much extracted what I needed.

Pro-tip: If you're designing a little multi-purpose computer module, and it has USB, and it has an expansion header... bring out a USB port on the doggone expansion header!  I'll probably be designing a piggyback board for this, and it'll need to transfer data, and the most convenient way to handle the data transfer (especially without messing with the system configuration) is via USB... but that means running a cable to the dang connector cluster.

Software note (news to follow)

If you build ReplicatorG from source - since there doesn't seem to be a prebuilt binary for Debian Squeeze, for example - you may find that the "model" tab has blankness where it seems a 3D view of the model ought to be.

Turns out this happens if you build it with openjdk 6.  Building with openjdk 7 gets the 3D preview, just where it ought to be.

Home ec

Wikipedia sez:

The term economics comes from the Ancient Greek οἰκονομία (oikonomia, "management of a household, administration") from οἶκος (oikos, "house") and νόμος (nomos, "custom" or "law"), hence "rules of the house(hold)".

Brigid has a nice post on the cost of real food prepared at home vs. "cheap" fast food.

Subject came up recently, here.  I've been making the occasional medium-sized semi-gourmet pizza, and Joy inquired about the cost of ingredients.  Hm.  Well... for vague guesstimates... crust costs maybe a quarter, probably quite a bit less.  Sauce, likewise.  Cheese, if it's mostly bulk mozzarella (i.e., high-protein hot-melt glue) with a little good stuff for flavor, something under $1.  Toppings, anywhere from a quarter to a buck and a half.

So, homemade pizza, putting no value on one's own labor, runs maybe $1.75 to $3.00 for a medium (serves one ravenous person, or two with moderate appetites), plus the cost of running the oven up to volcano heat, and wear & tear on the oven and the pizza stone.

Compare the price at your local pizza parlor, where you're paying for hired labor (and the associated non-wage overhead), rent on the business premises, depreciation on commercial cooking equipment, licensing, inspections, and presumably some profit for the owner (also the shareholders if it's a corporation, and figure in franchise fees if it's a chain).  Oh, and advertising.

This is particularly bad in high-overhead areas like Silicon Valley, but even in areas with more reasonable costs of doing business, fixing your own food is a lot cheaper than buying it prepared... unless you're bad at grocery shopping, or you could be using that food-prep time to earn money.

Hey, if cooking takes time away from your $50/hour job, by all means buy your food in ready-to-eat form and take advantage of the specialization of labor.  But if you have more time than money?  Learn to shop for ingredients, and to transform ingredients into meals.

It's amazing what you can do with rice, lentils, and some spices and such.

And remember: prior to the Industrial Revolution, the household was the basic unit of production.

Grrr.

Looks like my Subversion working copy of an important project has Gone Funny, somehow.

The repository is fine, as far as I can tell.  I can browse it with kdesvn, and nothing untoward happens.

If I try to view the status of one particular subdirectory in kdesvn, it hangs, waiting on non-interruptible I/O that never completes, so the only way to make kdesvn go away is to reboot.

If I point a shell at the root of the working copy and do "svn status", it looks reasonable.  If I've just crashed kdesvn, I may need to do "svn cleanup", which appears to complete OK.

If I do "svn commit" and fill in the comment, I get this:

Sending        Documents/Thing1/thing1-function-and-firmware.docx
Adding  (bin)  Documents/Thing2/field-reprogramming.odt
Adding         firmware/Thing2/Thing2-2013aug19.hex
Sending        firmware/Thing2/Makefile
Transmitting file data .

(Names changed just in case they might reveal anything.)

And I wait, and wait... what I've found on the Internets suggests that, eventually, it'll either complete or give an error message... for sufficiently hour-ish values of eventually.  It's been a bit over half an hour so far.

Update: a few hours later, still no go.  Narrowing it down: the problem has to be with either the new hex file or the Makefile.  The Makefile is open in jedit, but so's everything else.  The hex file is...

...wait a minute...

...is open for reading by a Windows program - a flash programming utility.  Which, despite (or due to?) keeping the file open, is noted for refusing to believe that another program could have been writing new versions (so it's actually necessary, if the hex file has been rebuilt, to select a different file, then re-select the updated one; merely exiting and restarting the program won't necessarily work).  Typical egocentric Windows program.

Anyway, terminating the Windows flash-programming program releases the SAMBA lock, which I guess allows Subversion to lock the file over NFS and complete the commit.