@Adnan Ok, so the entire project is mostly wrappers around C libraries. We (so far) support OpenSSL and commoncrypto on OS X. It's implemented as pluggable "backends" into the various interfaces. Here is the backend code for OpenSSL. github.com/pyca/cryptography/blob/master/cryptography/hazmat/…
I don't get the "but it can be up to 2^64 - 1 bits." part either. AFAIK the combined size of IV and block counter is always 16 bytes with AES-GCM. No diea how those huge IVs are supposed to work.
yes yes, that's right, the alien song from the Dr. Who episode "Rings of Akhaten". Two of my kids need to hear it to fall asleep. Sometimes multiple times.
@AviD The food truck notion is becoming more of a thing in the US, particularly in bigger cities. Meh, as long as everything is cooked to temperature and isn't rotting I'm pretty fine with it.
@AviD Naah, just tried a couple of directory names. Threw the s3 subdomain just because I saw once that slideshare served the files in a similar way a while ago
@RоryMcCune Oh I'm pretty sure they have a link to the file somewhere public on the site.
@CodesInChaos There is such a mechanism but it is like a custom MAC, then truncated to 12 bytes. It usually is best when the IV already has the right length.
> Dear Tyler Larson, Good news! You are entitled to a credit of $0.73 for some of your past Kindle book purchases. The credit results from legal settlements reached with publishers about the price of eBooks.
Was too tired yesterday, but today I deleted my "-1: Troll" comment and replaced it with something significantly more hurtful -- a well-reasoned and factually-accurate rebuttal.
@DavidFreitag I agree w/ @ThomasPornin -- if he's going to immediately delete comments that show he's wrong, then his blog isn't the appropriate place for discussion.
@ScottPack Here it is too cold for snow. -23⁰C this morning.
When air is below -12°C or so, it is too dry to allow snow to happen. However we can have "ice crystal" (condensation which results in tiny ice crystals in the air), which is beautiful.
Compose+^+0 yields '⁰', while Compose+0+'^' yields '°', which is another character. How wonderful.
@tylerl Nice, assuming I converted from °F to °C correctly, we had it like that about a week ago. It's a bit more April-ish now but still not too bad. Only the nights still go down to 5 °C so still no sleeping in the ditch :D
@ThomasPornin Coming from you, I actually believe you that it matters ... OK, American black bears. Their facial hair suggests they're white but I'm not sure.
Technically, the American black bear is faster on ground than the Asian black bear, because of its shorter claws; the Asian black bear has long claws for tree climbing.
An Asian black bear will run at about the same speed as Usain Bolt; an American black bear should reach 50 km/h or so. Polar bears have been rated at more than 60 km/h.
@TildalWave This was more a witty quote; however, it may matter because Asian black bear have a big patch of white fur on their chest, and fur colour may depend on local skin colour (it does so in cats, for instance -- that's a genetic mosaic effect).
@ScottPack Sprinting. Bears are lazy and won't run for long spontaneously.
Apparently, American black bear are somewhat beige, or tan.
@tylerl Well, it is a good question: "I am a werewolf, do you have cleaning tricks for linen ? Something about removing blood stains would be welcome, too."
It's not just about you. By forcing users to user TLS, they're creating a more secure environment for everyone.
Without TLS being strictly enforced, users are susceptible to attacks such as sslstrip. Essentially, making unencrypted connections an option leads to the possibility of forcing users ...
@tylerl Drat. I was going to edit that answer to just be No. with all the long drawn-out bits in an HTML comment (invisible to the casual observer), but your Markdown breaks that stuff.
@AviD After making the statement I realized that they're all three likely pretty equally likely and also non-blocking.
@tylerl Let's face it. The PCI-DSS sets a minimum bar for security controls. The fact that they're a minimum baseline, and still nearly impossible to reach, demonstrates how bad off we are as a whole.
@ScottPack It is an interesting point. PCI does seem to needlessly chase minutiae while ignoring huge swaths of attack surface. I guess when you design security constraints by committee, PCI is not far off from what you get.
> Currently, there is no good protection against the Man-in-the-middle-attack in NoSSL. The problem is that JavaScript is loaded from the server itself and run in the browser. Thus, the browser cannot independently verify the identity of the server.
yep the idea that the "complexity" of SSL should be relplaced with a javascript based sol. that doesn't even address one of the main risks just seems... odd
all that is both true and obvious. But I think they are actually aiming for a target with a diferent risk profile: e.g. your sister's cat's blog. and the main attacker is you, her 12 year old pesky kid brother.
so NoSSL would, hypotehtically, protect against that.
