Silicon Valley Redneck

Google sanitizing suggestions?

If you start typing "phn" into Google's search box, several suggestions pop up.

Most are of a Vietnamese character, though for some reason "pneumonia" is included.  Phneumonia?  None, though, is in Mad Arabish!

Adding the apostrophe gets "phn'glui" (which is a good start) and "phns" (which makes no sense).

Now, I could have sworn that, not so long ago, just "phn" got The Complete Phrase as one of the suggestions.

Did the Starry Wisdom Church, or perhaps the Alhazred estate, demand that the phrase be scrubbed?

Anyway, if you've ever wondered how "Phn'glui mglw'nafh Chthulhu R'lyeh wagn'nagl fhtagn" is pronounced, here's someone singing it.  Run!  Well, at least he didn't speak the name of H*stur.

Happy Gavrilo Princip Day!

Yes, today marks anniversaries of the beginning and the end of WWI.

Neither of which is exactly grounds for celebration, considering how they turned out.

I feel a great disruption in the supply chain

And another component turns out to be sold out at the moment....

Project Egbert uses one of the Atmel DataFlash (AT45DBxxxD) parts for FPGA configuration, a staging area for MCU firmware updates, and other handy things you can do with an SPI flash device.

Well, there are a couple of other projects in the works with somewhat similar requirements, and I figured it'd be a good idea to toss in the cheap little serial flash chip.

Oops.

The DataFlash parts are sold out.  All capacities, all packages, everywhere.  (Well, there's one obscure capacity/package combination that's in stock at a few places, but it's not one that makes sense here.)

There's a fair selection of 25-series flash chips available, and designing one of those in isn't a problem from a hardware perspective, but they're more of a pain to deal with (needing block erasure before writing, where the erasable block size seems to be much larger than the writable page size).  Also, it looks like different vendors' chips have annoyingly different handling of write operations, or maybe it's just that the datasheets are poorly written.  Anyway, using one of those parts would mean writing another driver for AGROS, as would skipping the vendor-specific SPI flash chips and just going for a micro-SD card.

'Course, I've been meaning to write the SD/MMC driver anyway.  Just not right now when everything's in crunch mode.

(Mouser semi-helpfully shows many thousands of the part I want as being on order, but doesn't report estimated delivery time.)

Oh, well... I can build a couple of extra copies of the Magic Programming Cable for the short term, teach the two or so other users how to use the programming app, and skip the staging memory.  For production, we may or may not want to support field updates anyway.

Bloody Stupid ergonomics

One might almost suspect B. S. Johnson of involvement in the new iPhone design....

Apparently the antenna design is such that, when one is using the gadget as a phone, one should always grasp it with the right hand.

Er... huh?

I almost always hold the phone in my left hand, for three reasons:

Firstly, I'm currently left-eared, owing to the right ear having suboptimal high-frequency response.  OK, so that one is just me.

Secondly, it sometimes happens that I need to take notes while I'm on the phone.  This involves holding a writing stick with my right hand, which is how most people write.  Thus, it's my left hand that's available for holding the phone.

Thirdly, if I've just been pushing buttons on the phone with my right hand (remember, most of us are right-handed), then the phone is already in my left hand.

Ergo: phones should be designed to be held in the left hand, to accommodate right-handed users.  Ambidextrous design is preferable, of course, but designing primarily for lefties can end up alienating the larger customer base.

Having a phone that doesn't work when held in the left hand is just so glaringly obviously wrong, you just have to wonder what goes on inside Apple.  Sinister conspiracy, or simple compound incompetence?

Another misinterpretable name

Is is just me, or does Droid-X sound like some sort of robot repellent?

If you have Droid-X brand robot repellent, do you still need Old Glory Robot Insurance?

...And, no, the Droid X isn't right at the top of my must-have list, though I haven't ruled out waiting until next month and getting one instead of looking for an Incredible right away (where "right away" is defined as "any time in the next three weeks").

Monday. Morning. Microsoft.

Among all the other stuff on the agenda for this week, I figured changing the connection from the office Windows box to the monitor might be a good idea.

See, the box in question was hooked up to the monitor with a longish VGA cable through a KVM switch.  Works OK for low-res displays, but at the monitor's native format of 1680x1050, it gets kinda blurry.  Which, naturally, is why the Linux box is connected using a DVI cable.

This had been OK-ish, as long as I was using Remote Desktop for most of the actual work on the Windows machine.  The resulting display had a few pixels shaved off the bottom to accommodate the KDE panel, but it worked fine for word processing and such.

Well, now I'm trying to do CAD work on the Windows machine, and it's a bit wonky with Remote Desktop - indeed, one of the CAD programs doesn't work with Remote Desktop at all.  Something about 3D.

So: the old All-In-Wonder Radeon 7500 card does have a DVI connector on the back.  Remove the DVI-to-VGA adapter, run a DVI-to-HDMI cable to the monitor's HDMI input, and away we go!

Or not.

Windows sees the monitor sometimes as a BenQ FP222WH, and sometimes as a "Plug and Play Monitor."  In either case, it thinks the maximum format is 1280x1024.

Which is, to use an esoteric technical term, wrong.

And it had been working, back when it was on the analog VGA connection.  It was a bit blurry, but it was running at the correct native format.

A spot of Googlage later, and I've downloaded a utility that can generate a corrected .inf file.  After much futzing, I've told Windows that that's the driver for this monitor.  A reboot later, and Windows consistently has the right name for the monitor.

It does not, however, have the correct resolution.  It offers 1280x1024.

ATI's "Catalyst" utility lets me kick the format up a little, but nowhere near what it should be.

So, what should have been a half-hour expedition among the cables and dust wound up eating the entire morning, and it still isn't working right.

Ain't plug'n'pray wonderful?

And, y'know, I didn't have this problem with Linux.  Linux just works.

Enough venting, and more than enough time spent on the stupid Windows problem.  Time to get some actual work done.  Then, later, I can return to the ferschlugginer Windows issue.  (Probable solution: "buy a new computer to go with your new cable.")

Update: I swapped the cables.

So now the Linux box is connected to the HDMI port on the monitor, and the Windows box uses the DVI-to-DVI cable.

And, whaddaya know: Windows wakes up using the old display setting (1280x1024), but the monitor presents it in 4:3 format with a little black band at the bottom, rather than stretched to fit the screen.  The display settings thingie now lets me choose 1680x1050, and all's well.

Returning to the Linux side of things: it looks happy so far, except that for some reason a background color of ArrGeeBee*(95%,95%,95%) comes out as a series of pale-blue bars on light gray, instead of uniform light gray.  Huh?  Changing the 95%s to 96% gives proper light gray, but one of the colors in this here Typepad composition page has the same banding effect.

Also, the Xorg configuration utility seems to have set things up for a generic plug-n-pray monitor, so if I reboot I'll presumably get stretched 1280x1024, unless I go in and manually set the display format.

Looks like the monitor has Issues on the HDMI port, which I'd always thought was just supposed to be a sort of updated and shrunken DVI.

Maybe I need to buy a DVI monitor switch.  Oog: pricey, last I checked.  Well, I need to make a computer-store run soon anyway, on account of yet another fan starting to get noisy, so maybe I can look at switches.  And there's always Newegg.

* Bypassing Typepad's helpful translation of the expression as I'd originally typed it.

Update 2: picked up a 2-port DVI KVM+audio switch (IOgear GCS932UB).  Kinda overpriced despite being a returned unit (yup, Fry's), but today, while ordering one from Amazon or B&H would get it to me next week sometime.

Anyway, the problem is solved, and the Windows box can now be connected to the speakers, should I ever feel the need to do such a thing.

One thing IOgear doesn't mention in the little installation booklet is how the gadget is powered.  Turns out that applying power to any of the four USB plugs does the trick.  I'm not actually using the K nor M switch functions, what with the gizmo being of the USB persuasion, and me having a PS/2 keyboard and mouse already routed by way of a 4-port KVM switch.

So now I've got a little button on my desk to switch the input to the monitor, independently of switching the keyboard, mouse, and VGA.  I was having to use a button anyway, and this is easier to find than the little DVI->HDMI->VGA button on the side of the monitor.

And the display is looking fine, from either machine.  I do need to rearrange the icons on the Windows desktop, though, so's they don't get auto-rearranged when switching between the 1680x1050 direct connection and the 1670x975 remote desktop.

Being the early late adopter

One of the annoyances of doing business these days - actually, the last decade and a half or so - is MS Word.  Specifically, the way the default document format keeps changing, so if you're exchanging documents with clients and such, you pretty much need to have the latest version.

I'd been getting by with a combination of OpenOffice.org and Office 2003: O2k3 for documents that arrived in .doc format and needed to be edited and returned in that format without the formatting getting disrupted, and OOo for everything else, including reading the occasional document that showed up in .docx format.

Well... one of those Office 2007 users had started sending .docx files that crashed OOo, leaving me waiting for the boss to run them through O2k7 to make them readable.  This wasn't good.

And so it came to pass that, on this, the second day of its availability, I went forth and procured a copy of Office 2010 Home and Business.  Yeah, the business-licensed version, since it is for business.  And, being as how it's exclusively for business use, it's a business expense.

So now it's my turn to be the annoying guy who forgets to save documents in a backward-compatible format!

And, one of these days, I may even get a smartphone.  Not this week, though.  Got a crunch schedule, and I just blew a day and a half tracking down a horrid bug that turned out to be an offset that was out of sync between an assembly module and a header file (which puts all that time solidly in the non-billable category).

Back to the crunch now.  All three of them.

Microsoft strikes again!

When writing a little script that returns dynamic content for your AJAX webapp, remember this:

You must include the "Cache-Control: no-cache" header if it's gonna be used with Explorer.

'Cause IE8 thinks it's a dandy idea to cache responses to XMLHttpRequest.  Even if the URI has a clue to dynamicality, like one of those question-mark thingies used to set off parameters passed to scripts.  And if it's got a response cached, there's no point making the request again.  Which makes the dynamic content not work, and also makes the control functions only work once each.

Now that that's fixed, the toaster's embedded web interface works with all the modern-ish browsers that I have handy.  Except for the minor detail that setting the background color of table cells in JavaScript doesn't seem to work in IE, so the indicator lights are out of commission.  One more ferschlugginer detail....

Update: heeeere we go.  Seems the bgcolor attribute has been deprecated into semi-oblivion in IE8.  If you set it in the HTML, it works.  If you set it in JavaScript, with

e.setAttribute("bgcolor", b.color);

it changes the bgcolor attribute displayed by the Developer Tools, but doesn't actually change the displayed color.

What actually works (tested in IE8 and Firefox, so far) is

e.style.backgroundColor = b.color;

Update 2: huh.  Now it's not working with Konqueror.  I thought that used to work... but now that it's working in IE, FF, and Galeon, Konqueror is completely failing to update its display (though the function buttons are sending the applicable commands to the toaster).  Oh, well: the customer probably isn't using Konqueror, so sorting it out can wait until some other day.

Well, somebody's having a good opening week.

Out running errands this afternoon (catching up on shopping I hadn't gotten done while I was home sick), I decided to check out the new Sprouts market that just opened this week, down there on El Camino in Sunnyvale.

I used to do most of my grocery shopping there, back when it was Nob Hill, before it was Circuit City*.

First impression: the parking lot was crazy-full.  I ended up parking near the far end, at the office-supply and sporting-goods end of the strip mall.

Having gotten parked... well, the store was crowded.  Supplies were keeping up with demand, though; they didn't appear to be in danger of running out of anything.

I'd been vaguely expecting something Whole Foods-ish.  And, if the reference point is the Whole Foods in Palo Alto in the early 90s, that'd be about right, but it's a lot smaller than the new ones.  Somewhere between Whole Foods and Trader Joe's.

Plenty of good stuff there, in the areas I checked (cheese, bakery, meat & poultry, produce).  Prices overall seem reasonable, and some of the weekly specials are darn cheap.  Quality seems good, as far as I could tell, but the proof of the pudding and all that, and so far I've only tested the bag of cherries I bought there, which proved tasty.  Chicken to be tested a bit later.

No express checkout.  Maybe they'll designate one once the mobs dissipate a bit.  Traffic patterns approaching checkout are suboptimal when the store is crowded.

Oh, and the web site is dysfunctional, at least in Firefox.  Shiny flash stuff, with a link for a non-flash weekly ad, which gets a bunch of decorations but no content in FF.  (Ah, but in IE it gets a store chooser, which leads to a choice of flyers.  So it's a Windows-only site, then?)

I expect to be adding it to my list of places to shop for groceries, once the new-store madness dies down.  (Also, I normally make my grocery run around 0930, after morning walkies, and maybe it wouldn't be so busy then.)

* I hardly ever shopped for groceries there while it was Circuit City.

Ferschlugginer bushy-tailed tree rats!

Not only have the squirrels been digging holes all around my yard, eating seeds they happen to dig up, uprooting small plants, and (rattles worry balls) stealing my strawberries....

Somebody, I say somebody, gnawed about a third of the little branches off one of my recently-planted blueberry bushes.  Right at ground level.  And left 'em lying around on the ground.

And I don't think it was beavers.

A small yappy dog would probably deal with the squirrel infestation, but then I'd have a small yappy dog infestation, which isn't necessarily an improvement.

I guess I need to deploy chicken wire as a temporary measure, and maybe consider putting something unpleasant along the tops of the fences.  Stainless steel wool, perhaps?  Little bits of nitrogen triiodide blotter?

Long term, maybe a yard dalek?  Or an automated yard-defense pea-shooter with millimeter-radar targeting?

Check yer filters.

Yeah, I know.  You've heard that line before.  Maybe six months ago, when the telephone solicitor insisted that you need to have your furnace filters professionally inspected every fall, even if your furnace doesn't have a filter, nor a place for one to be.

Well... what with summer weather showing up... now it's time to check the filters on your computers.

This morning, the system log showed that Perdita (the dual-core Linux workstation) was having hot flashes.  So: power down, and have a look.  Ugh!  Lotsa dust.  Also lotsa heat.  Fans all spin freely, so the problem must be impaired air circulation.  Perdita and Vetinari (the FreeBSD server) both inhabit Cooler Master Centurion cases with the filter fronts... and both had their filters pretty well saturated with dust.

Perdita's CPU is now sitting at around 55°C, as compared to 97°C before a session with the air hose.

Don't have temperature measurements readily available for Vetinari's innards, but some weekend soon I need to declare a server holiday for interior dust removal, UPS reconfiguration, and also swapping a more vigorous fan into the thermal defenestrator, which is still allowing far to much heat to build up in the office.

Freakin' hilarious!

Killing time while waiting for software to load, and finding the ROVcams a bit boring at the moment, I happened across We Con the World.  Go watch.

(Via UJ @ Blackfive.)

Why not nuke it?

Yes, the question is being asked a lot, as BP's well continues spewing oil into the Gulf.

It's even made it onto the front page of the morning fishwrap.

The folks at TOD have pointed out several perfectly good reasons it's a dumb idea, and also raised a not-so-good objection or two.

Let's start with a non-reason it wouldn't work: none of our nukes are designed to operate under 5000 feet (or 18000 feet) of water.  Well... perhaps not, but it wouldn't be too difficult to put an SADM in a high-pressure, watertight capsule specially built for the purpose.  Assuming, of course, that anyone could come up with an SADM.  Presumably there's something in the arsenal that could be adapted to the purpose.

On to the real reasons it's a dumb idea!

Consider the objective.  Presumably, you're trying to collapse the wellbore and fill it with broken rock.  Except... the pressure in the reservoir is high enough that this likely wouldn't work; a BOTE calculation suggests that the rubble would just get blown out, unless (a) it got stuck in the BOP, and (b) the BOP and upper casing remained in place - neither of which I'd care to bet on.

On to methods.  Obviously, just placing a nuke on the seafloor next to the well and setting it off is something that could be done quickly (timescale of a week or so, unless you count the time to get permission, in which case figure several years).  And what effect will this have, exactly?  Hmmm... it'll stir up the silt something fierce.  And certainly remove the BOP, the upper casing, and any hope of ever containing the flow while waiting for the relief well.  None of these sound especially useful.

That leaves shoving a nuke down the relief well.  Which means waiting for the relief well to be drilled, which puts us in August anyway... using a nuke would save the last few feet of precision drilling and allow the first "somewhere in the ballpark" attempt to be used.  But!  There's the minor detail that broken rock seems unlikely to contain the reservoir pressure, and there's the potential for blowing the entire top off the rock formation and releasing the reservoir contents in one fell swoop.

If we want to put a nuke at the base of the well, where it might to some good (or, more likely, a whole lot of harm), we still need to wait for the relief well, unless we use Project Vulcan.

The Daleks also have underground-nuke technology, but they're not due to arrive for 140 years yet.

And why would we want to depend on villains, anyway?  Has anyone tried calling International Rescue?  Between Thunderbird 4 and the Mole, they ought to be able to deal with this, but they don't come all by themselves, you know; you have to say their name on the radio.

Return of The Craw!

Well, it seems that BP has given up on using that diamond saw to get a nice clean cut on the lower end of the riser, and is instead just going to snip it off.

Could get Interesting, though I guess the "crimping leads to excess pressure leads to a blowout down below" menace is mitigated by the existence of that slot that the diamond saw did manage to cut.

Sometimes my job looks easy.

Like, by comparison to the crews working in the Gulf right now....

Just had to drag myself away from the chicken soup and head over to a client site for a problem-solving session with a rather tightly-wound tech who has a rapidly-approaching deadline.

And I thought: hey, it could be worse.  There could be millions of people watching on a live webcam.  And the equipment could be malfunctioning in a visually interesting way - say, copious quantities of oil spewing forth - rather than just not sequencing a couple of signals on at the expected times.

After about an hour and a half, all problems had been resolved by either removing the test from the checklist or documenting the behavior - what we call putting a necktie on the bug and calling it a feature, except that these features were behaving as designed, and it was the documentation that was out of sync.  Really.

And now the shipment can happen on schedule, and I can take a nap.  Tomorrow, I gotta get productive on my own work again, before I find myself on the critical path.

Well, look who's going digital!

Watching the ROVs at work beneath the Gulf, I keep thinking that this calls for a parody of "The ROVing Repairman-O."

This leads to Googling.

Which leads to the discovery that Annie Lore now has a (one-page) web site, and a CD version of The Grey Cocked Hat in the works!

(I had a CD of that, but it was in the CD player in my Jeep when the player was stolen a few years back, and I haven't gotten around to re-digitizing from the tape, which is still safe in my office closet.  Perhaps the CD will come out before I get around to it.)

The wretched news handling continues

Being still home with the miserable tail-end of last week's cold, too sleep-deprived to do anything useful and yet too hopped up on pseudoephedrine to sleep, I've spent an unreasonable part of the day watching the BP-BOP channel.  (Beepy-bop?  Bippy-bop? Bippity boppity boo?)

Well, this morning The Craw arrived, and chewed on the riser a bit, then (while I was looking away) vanished from the scene, leaving the riser dented but uncut.  Word on TOD was that the Super Snippy Shears had withdrawn in disgrace after failing to perform the scheduled riserectomy.

Just about then, word started making the rounds - breaking news, this - that BP had abandoned efforts to stop the flow of oil from the top, and that it wouldn't be stopped until the relief well could be completed in August.

Eek?

Well, in fact this had nothing to do with any difficulty in the riser-cutting operation.  It was well-aged news that top-kill, with or without the junk shot, had proven futile, and that the LMRP option would be pursued in order to contain the escaping oil until the well could be stopped permanently.  Not at all new news that today's containment effort had been abandoned.

You'd think that a competent PR department would make a point of emphasizing the times of press releases, and exactly what had changed and what was continuing.

Sure enough, Mr. Craw reappeared a bit later and snipped through the riser, and now the whirly-wire thingamajig is sawing away at the riser just above the flange.  Oil started spewing from the cut a while ago, indicating that the wire had made it through the wall of the riser.  (The presence of oil at that point doesn't prove anything about the path it took to get through the BOP; the drill pipe is probably broken at the kink, and may also have been punctured by the shear rams.  The brief interval between completion of sawing and arrival of Mr. Elmer P. Capp may be more informative.)

Hm.  The saw had stopped for a while, with a jet of oil in its face.  Now it's running again.

Not planning to liveblog this; for group liveblogging, see the current thread at TOD.