Jump to content

Wikipedia:Reference desk/Archives/Computing/2018 June 12

From Wikipedia, the free encyclopedia
Computing desk
< June 11 << May | June | Jul >> June 13 >
Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.


June 12

[edit]

Virtual "RAID 1" in flash memory by splitting into equal sections?

[edit]

I would like to store a small amount of important data on some USB flash drives or microSD cards. Since parts of memory can sometimes fail and lead to loss of data, I was wondering whether there is some kind of software that makes a sort of virtual RAID 1 on a single drive by splitting it in half or even in thirds and then have two or three copies of the same data in the different sections so that if the memory should fail in one section there would be others. Of course I could just copy everthing two or three times but that's a lot more manual effort to maintain. — Preceding unsigned comment added by 185.230.100.66 (talk) 19:28, 12 June 2018 (UTC)[reply]

You could certainly partition the flash memory and use software RAID to assemble the partitions into a data volume. It's doubtful that there would be any benefit.
  • the extra writes needed for maintaining a triple RAID-1 or a RAID-5 will greatly increase the wear rate on the flash memory. (Anything less will not let you know which copy is intact if there are only two copies and they disagree.)
  • failure modes in which the whole device becomes unreadable or largely corrupted are not protected against and will probably dominate any data loss risk calculation (you probably can't control how the device firmware co-locates data as part of its wear-leveling algorithm).
The "I" in RAID stands for "independent"†. If the devices are not sufficiently independent, there's very little protection. — jmcgnh(talk) (contribs) 17:40, 13 June 2018 (UTC)[reply]
† yes, originally it stood for "inexpensive".

Christmas coming early?

[edit]

If your bank receives through the post a cheque made payable to you it will ensure your name appears on it before taking the money from the sender's account. If it receives the money electronically it does not check that your name appears on the payment instruction before taking the money from the sender's account. This means that if the payee is named, for example, as "Santa Claus" you still get the money provided the sort code and account number are yours. When the payer queries the payment ("Did you credit Santa Claus as I directed?") it refuses to say whose account was credited. The payment is subject to the bank's terms and conditions. What do these terms have to say about this? 86.131.233.241 (talk) 19:38, 12 June 2018 (UTC)[reply]

Earlier this year the Payment Systems Regulator initiated a consultation about this which remains open to September. A sample T&Cs I looked at reads:

When you wrongly identify the person and/or account (including account number and/or sort code) to which an Electronic payment should be made, we will make reasonable efforts to recover the payment made, but we will not be liable for any losses which may be incurred by you.

The banks have announced the introduction of a "confirmation of payee" scheme under which the sender of the money will be asked to confirm the name attached to the recipient account. As things stand at the moment, the bank will not effect the payment unless the account name is provided. What changes are planned here, and will they result in compensation for victims of APP (Authorised Push Payment) fraud, where a fake email, ostensibly from a creditor, changes the genuine account details to those of the fraudster? 86.132.186.246 (talk) 19:16, 13 June 2018 (UTC)[reply]

The banks usually refuse compensation where the account holder has voluntarily made a payment, regardless of the nature of the faking. It is possible to set up a fake duplicate name as well as account, though the checks by British banks on setting up new accounts are quite onerous, and a fraudster would need lots of information on the genuine account holder. The usual advice is to see the cheque book or paying-in slip of the destination account before making a new payment, or to transfer £1 and check that it has been received by the correct account. Never make payments on the basis of an e-mail. This is just common sense, not legal advice. The confirmation of payee facility will help, and is long overdue, but will not completely solve the problem. The most common fraud is via international bank transfers. Dbfirs 10:50, 18 June 2018 (UTC)[reply]
I'm still unclear as to what protocol is being proposed here. Will the recipient bank, on receipt of the payment, notify the payer of the name of the account holder in every case where a different name appears in its records and await confirmation of the payment before crediting its customer? 86.132.186.246 (talk) 12:55, 19 June 2018 (UTC)[reply]
I think both protocols will be implemented. For a business account, the details are public, so the bank can disclose the name and company registration number for the person transferring the money to check. In the case of personal accounts, the account name would be subject to data protection laws, so the bank would just confirm a match or otherwise. Some banks are hoping to start implementing this by the end of this year. Dbfirs 19:03, 19 June 2018 (UTC)[reply]

Do you mind this Internet cookie? How about this one? And this one? And this one?

[edit]

I'm fed up to the back teeth of websites asking my permission to put cookie on my PC. My idea to comply with the stupid cookie regulations is that there should be a browser setting whereby a user can automatically tell websites that they're fine with cookies just like they were for a decade before these rules were implemented. The website then checks this setting before asking with another intrusive message or banner. To whom do I suggest this? The World Wide Web Consortium? It requires co-operation by web designers and web browser developers. I'd preferably like it implemented about three years ago... --185.230.100.66 (talk) 22:16, 12 June 2018 (UTC)[reply]

That sounds similar to Do Not Track. W3C would probably be your best bet, since they have a hand in supporting DNT, though I don't know if there'd be much support for opting in to tracking as opposed to opting out. Advertisers would probably like it more than DNT, though. clpo13(talk) 22:31, 12 June 2018 (UTC)[reply]
There's a bit mentioned at HTTP cookie#EU cookie directive that sounds similar to what you're suggesting; P3P would have sent privacy policy information to the browser via an HTTP header and set cookies according to the browser settings. It looks like it was never very widely supported, though. clpo13(talk) 22:41, 12 June 2018 (UTC)[reply]