De-duplication doesn't work that way. The system tracks duplicates at block level and if a configured cluster size of blocks is identical, the file block stream is chained through a single copy of the blocks. If a new version of the file appears, it will fail the block check and will have space allocated to it. If more copies of the new version then appear, they will be chained through the single copy of the new version.

If you delete a single instance of a de-duped file, it is handled the same way as multiple hard links to a file on a unix file system. The FAT entry (if you don't mind the archaic reference) is removed and the reference count to the data is decreased by one. No blocks are freed for overwrite if other FAT entries reference the blocks. So NO, an installer deleting one instance of a de-duped DLL will not remove the contents of the file from disk as the blocks are referenced by other files.

The transcriptions I've seen have all maintained the phonetics of the Arabic word al-qÄÊidah as al-KY-dÉ(TM)

What we have here is

al-kwid-a

, as in quid pro quo or the slang for a British pound.

They didn't spell Al-Qaeda correctly at all. That looks like the spelling of someone who's only ever heard the name and spelled it the way they were taught English.

Or part of the development and testing process for a new generation space suit?

If you could point us to the source code of the Windows services manager engine, we'd appreciate it. :)

The problem that the Onion Pi is not granny-friendly. The problem that the Onion Pi needs to be assembled and requires RP Linux knowledge to set up in the first place.

Some people want the challenge of making the device, others just want to plug it in and go. That's where this comes into play.

They had similar levels of data collection on the tech previews and betas of 7 and 8, and switched it off on the RTMs. I don't think even MS has the desire and capacity to collect that much data from a billion computers.

At least you can diagnose and fix issues with shell scripts with vi and a bit of knowledge. Try that with a binary blob that stores its data in a binary store.

Either yours or his is the correct analogy. If the images come from the Snapchat server, then they are not deleting their images as they claim they are. That is your analogy.

If the prevailing theory that the popular 3rd party app for Snapchat is breaking the delete-after-x-hours promise by uploading the image to a non-Snapchat server so it can be accessed later, or uploading the image without knowledge of the users of the app because they are sick buggers who want to see what the pics are (NSA I'm watching you), then his analogy is the better fit.

Either way, people stop assuming your arse and tits photos are secured when you trust the cloud.

Link to a proper article about it. http://news.agu.org/press-release/satellite-data-shows-u-s-methane-hot-spot-bigger-than-expected/

Because it is a feedstock into their polio vaccine production line.

Might I ask what specs/OS you're running? Firefox on Windows 7 on an i5 didn't even hesitate with the page.

I think the difference here is the same as the difference between buying your own delivery trucks from Mack and contracting FedEx to deliver your products from factory to retail store.

The earlier days of space flight were like buying the various bulk components of a truck, engine from GM, chassis from one metal shop, driver's cabin from another, electronics from Lucas and then building the full truck up. Come the era of the shuttle, the delivery truck came fully assembled from Mack, but you still have to pump the diesel, change the tires, load the cargo into the back yourself.

Now it looks like calling FedEx and telling them you've got fifty packages in London and you need them in Bogota by Sunday. They pick it up, containerize it and ship it.

Hope that helps?

I think the companies knew that multiple contracts were an option, but judging from the press coverage in the week or two leading up to it, the business pundits thought it would be the traditional multiple proposals, one winner option.

I don't know what will happen if both deliver a fully functional vehicle by 2017 within the budget. We still could be in the Y stage of the process, e.g. the YF-22 and YF-23 fighter technology demonstrators which were functional aircraft leading to the choice of the YF-22 to go into production as the F-22.

NASA could also go head post-2017 with both companies supplying vehicles and launches. We'll have to see.

Not Boeing alone, and not SpaceX alone. This is the best possible outcome for NASA, not reliant on a single supplier like before.

The fact that to deliver the same development and certification process costs $1.6 billion less for SpaceX over Boeing is also interesting. Some are already saying that it is a bigger win for Boeing and that SpaceX is a backup plan, but since the amounts are what the two companies bid on the project, it shows how economical SpaceX believes they can be.

And that there are two companies still competing should reduce the risk of deliberate cost-overruns and delays. If one can get to full certification a year or more ahead of the other, it will be a huge blow to the second-place finisher's chances to win the final operational contract.