SpaceX is a new company and had 19 launches before failure on this. Even now, we still do not know what caused this.

SpaceX has had more then one failure, this is just their first failure on the Falcon 9 series that resulted in loss of all payloads. The Falcon 1 rocket had 3 failures out of five launches, but was considered a "test" project. Falcon 9 launch #4 was a partial failure where the secondary payload failed to reach orbit. Launch #7 was an almost failure due to a fire during flight down in the Octaweb engine area.

I believe there has been at least one launch where a first stage engine acted up, but I can't find the reference.

I've used a lot of keyboard nubbies over the years. The first one (c1995), the button was like a joystick and down by the spacebar. That one sucked. The next units were Toshiba Tecra series laptops with the nubby between the g/h/b keys. For the last 8 years, I've only used Thinkpad T series laptops. Both the Tecra and Thinkpad nubbies were quite good for the purpose.

One trick with pointer nubbies is that you really need to turn up the mouse pointer movement sensitivity to maximum. You'll overshoot on movement for the first week, but then your index finger will thank you because you need less effort to hit a target. Again, the purpose of the nubby is not to replace the mouse, but to let you do 90% of point-and-click operations (clicking buttons, positioning the cursor, basic drag-n-drop) without taking your hands off the home row.

If you're not a touch-typist, you won't see the benefit of a pointer nubby. If you do a *lot* of copy/paste or complex mouse operations, then a regular external mouse is better.

My current work unit is a Thinkpad T540p -- on that one, I dislike it, not because of the nubby, but because there are no physical left/middle/right mouse buttons. They got subsumed into the touchpad click surface. Fortunately, for the T450 and T550 series, they have brought back the physical button below the spacebar.

Hmm, that's a good point. However if too many people were going directly to the root servers, eventually wouldn't they take some action to limit access to whoever needs it (as opposed to who wants it) to reduce the workload on the servers?

The only reason BIND / unbound talk to the root servers is to find out which DNS servers are authoritative for the various TLDs. The DNS root servers do not return the answer for "what is the IP address of maps.google.com", they only return the answer for "what DNS server is authoritative for .com?". Once your DNS server has the answer for ".com", it goes and asks the ".com" servers about what server handles "google.com".

I've read that a well behaved DNS server will only talk to the root servers about once every 48 hours, or whenever it hits a new TLD that is not yet cached.

Services like DNS really belongs at the network level, not the local PC level. If only for the possibility that there are 2+ people on the local network who query the same thing and the DNS server can cache / return the results. Or, since the network server is likely to be left on 24x7, it can cache answers across reboots of your local PC/laptop.

Something like pfSense on the firewall to the outside world with "unbound" running does just fine for this. You can configure it to talk to your ISP's DNS servers, Google's servers, or set it up to start at the root DNS servers and do its own heavy lifting.

Google maps also has the birds eye view now, with four different perspectives. They even 3D model the buildings (very roughly) and foliage.

Not sure if it's everywhere, or just in select cities, or only on faster PCs.

Specifically, Samba 4.2 with Snapper. It's probably still a scheduled snapshot, but it looks to be better then how we did things in 4.0 and 4.1.

I dropped Firefox because they no longer have a usable sync across multiple devices. With two laptops, a desktop, a cell phone and a tablet, some sort of bookmark/password/history sync is absolutely essential to me.

Right now my options are... Chrome.

The killer feature for Firefox or Opera would be to offer some way to sync to any WebDAV backend. Then I could setup something like Owncloud / Seafile on my own hardware..

Flash memory has, historically, been very expensive which is why you don't see larger amounts. There's also the long lead time and certifications for any new product. That's probably two years, minimum, between initial spec and reaching end users.

Not to mention that you're talking about low-end phones, which are always designed to hit the minimum specs. If you want bigger/faster, then you need to pony up for phones like the iPhone6 or Galaxy S6 which come with 64GB and 128GB options.

One script I use before putting a drive away for a very long time is to use Cygwin and sha256sum on the entire drive. When I pull the drive back out of storage, I can run another script to validate that none of the files have changed and that all sectors are still readable.

(laughs) I've actually been prepping a pair of USB drives for offsite backup since Friday. It takes 2-4 days to run "badblocks" with three clean passes, then another day to run "shred" on the drive. If it passes that burn-in test, then the drive is generally good to go for a few years of service.

You just reminded me that my shred pass was finished and that I should finish setting up LUKS encryption and add them to the backup pool.

My backups are all written to a central file server, which has a 4-bay USB enclosure attached. So once per day, the server copies the backup files off to one of the four USB enclosure drives. Then I also have a pool of external USB drives that get carted offsite semi-frequently to a safe-deposit box. Then there's the annual backup to a trio of SD cards, kept in the safe.

At the last company, we had six generations of external USB drives that went offsite each week. That may not sound like a long retention period, but using rdiff-backup or attic-backup, each drive had incrementals going back 27 or 54 weeks. I'm pushing them to ramp up to eight generations.

For my primary small business account, I went with KolabNow which has pretty good spam filtering. The downside is that you're going to spend $65-$95 per year on the service depending on your mailbox size. They need to cut their prices a bit or offer more storage for what they charge. They have more of an incentive to stop spam as email is their primary business.

Running your own mail server is not worth it unless you absolutely need control over storage of the mailboxes for legal reasons. I've admin'd an Postfix/Dovecot/SA for the last decade. Setup takes 3-4 days, you have to tune it monthly for the first year, and then keep an eye on it. I estimated that we blocked about 90-95% of all inbound spam at the firewall, without too many false positives. The client-side spam filters took care of the rest.

That requires DNSSEC and DANE to be effective. There's momentum for both, but neither will hit mainstream until Google's Chrome forces it.

Ultimately, I expect a mix of pinned-certificates, DNSSEC/DANE, and cloud-based reputation for certificates (is everyone else seeing the same certificate?).

Key management is hard -- really hard. It's the weak link of modern encryption.

Having played both sides of the fence, Java vs C# for server-side stuff is about equal. Especially if you use a good framework, inversion of control, and unit/integration testing. At this point, you can succeed with either.

The main downside right now with C# is your limitation to running on top of Microsoft Windows/Azure servers stack. Support for running against non-Microsoft technologies (such as PostgreSQL, or under a different O/S) is still a rough edge.

#3 is pure shill. MSDN documentation is crap for 90% of what you search for compared to how it was back in 2000.

The big problem with MSDN is that they change URL schemes every 2-3 years, breaking every reference URL that you might have saved. Then there's the almost, but not quite, completely useless form of the documentation which tells you everything you already know without making the water less murky.

I prefer one GPG file per site. Downside is that it exposes the site name, but also means I only decrypt only a single site password at a time.

Bonus points for putting the files into a version control system (git/svn/hg) so that you can cleanly sync them between PCs.

And making backup copies is as easy as stuffing the ASCII armored block into an email. Or printing it out for OCR'ing later...