Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
pierrec
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
54 ms
·
451.
▲
by
pierrec
10y ago
>What common features that the user needs on the web rely on it? - Website owners and content creators use the Referer to see where their users are coming from, which allows them to make content which is more suitable, which indirectly
452.
▲
by
pierrec
10y ago
Yes, good point, and it was easy to verify. Disable plugin; search "what is my referer"; click on first result: Your HTTP referer is google.ca . Enable plugin; repeat: Your HTTP referer is No referer / hidden .
453.
▲
by
pierrec
10y ago
After some testing, it appears that this plugin offers better referer hiding than Google's redirect. I've verified that it completely hides the referer, whereas Google at least reveals that you're coming from Google (which
454.
▲
by
pierrec
10y ago
This seems to work pretty well on Google search results. If you haven't noticed, Google search has this anti-feature bordering on the dark pattern where hovering a link shows the correct destination URL but right-clicking it (eg. to co
455.
▲
by
pierrec
10y ago
I was curious about how this worked so I looked into the source a little (even though Haskell isn't exactly my cup of tea), and WOW... This is just amazing. The most important part of the source is highly educative literate haskell: h
456.
▲
Nominative determinism in hospital medicine (Limb, Limb, Limb and Limb, 2015)
(publishing.rcseng.ac.uk)
2 points
by
pierrec
10y ago
|
0 comments
457.
▲
by
pierrec
10y ago
My quick-n-dirty benchmark for measuring the exhaustiveness of a stock photo website is how many photos I can find of man laughing alone with fruit salad . One of the staples of corporate promotional photography. In this case, 0. Without l
458.
▲
by
pierrec
10y ago
>I wonder why they didn't just use a test clip like this one Because they tried and it didn't work (third post). It seems to depend on the mobo as well as the programmer you're using, but sometimes SPI programming without
459.
▲
by
pierrec
10y ago
It's kind of a gray area, but looking closely I find it leans heavily towards the buggy, "unexpected behavior" side of things. Here's why: when encountering HTTP->HTTP redirects, wget already ignores the redirected fi
460.
▲
by
pierrec
10y ago
Yeah, I'm writing this without having first-hand experience regarding the current performance of Windows 10 as a guest. Apparently VMWare has good support for it, but searching for reviews on the actual performance only turns up a bunc
461.
▲
by
pierrec
10y ago
>you can't run Windows natively that way, it has to be a guest operating system. Then you need multiple videocards to pass through, and the guest operating systems require custom drivers at that. I, for one, think that sounds excell
462.
▲
by
pierrec
10y ago
That was also my first thought, but after consideration, there's more to "mise en abyme" than simply placing a story within another. I think in France everyone learns this term at least in high school, specifically applied to
463.
▲
by
pierrec
10y ago
Meh. "it was trivial to discover with a dumb fuzzer, and it's surprising that such a bug could even survive until 2015, with so much work being supposedly put into the security of font processing." (Emphasis mine.) I suppos
464.
▲
by
pierrec
10y ago
Regarding the question of whether browsers sanitize the fonts, I'm looking at the chromium issue linked to what the article calls a "weaponized exploit", and even though it's marked as fixed (obviously), it's not cl
465.
▲
by
pierrec
10y ago
Being not entirely up-to-date on current browser security, what's the risk of a webpage just including a CSS: @font-face { src: url("/fonts/evil-font.ttf"); ... } And making use of these? Do current browser
466.
▲
by
pierrec
10y ago
Also RTFAing, the authors express their surprise at how easy some of the vulns/crashes were to find using very basic fuzzing. So while @x0x0 might be jumping to conclusions a bit harshly, saying that MS did little to no fuzzing on thei
467.
▲
by
pierrec
10y ago
In short, " floating particles can travel upstream as much as 1 cm up a waterfall and several metres up a channel against a downstream fluid flow. " (this is from the actual paper and not even mentioned in TFA despite it being the
468.
▲
by
pierrec
10y ago
You're still missing one... https://www.erudit.org/revue/ttr/2010/v23/n2/1009160ar.html (While not 100% about the original version of Nitobe's book, they still analyze it and provide a bun
469.
▲
by
pierrec
10y ago
While it might be a more authoritative place to start, it's not the only academic paper in the sources, and it's definitely not the entire argument undergirding this post. I for one appreciate when he pulls out a citation that mak
470.
▲
by
pierrec
10y ago
I was immediately very interested by one of the references, which extensively compares five different versions of Nitobe's book: the original (English) and 4 translations (in Spanish, French and Japanese). They study the overall texts
471.
▲
by
pierrec
10y ago
It's a ridiculous practice and storing the passwords in plaintext is deplorable. Banks have your phone number, (which can't simply be changed online) and that is the best additional line of security. The scrambled on-screen keyb
472.
▲
by
pierrec
10y ago
Because there are still people, such as myself, that occasionally use heavily metered connections, and the percentage of people to whom this applies will not drop to zero in any foreseeable future. So I suppose the real question is: even th
473.
▲
by
pierrec
10y ago
Skimming through the docs, I'm still not sure how they can guarantee that an enclave can't be meaningfully emulated. Do they simply mean that you can't emulate a specific processor (since each processor uses a unique key)? Th
474.
▲
by
pierrec
10y ago
>Seems like a good assumption So, they're making an OS for the remote future? A laudable endeavor! In the meantime, I'll stick with an OS made for the present. (posted over a temporary LTE connection where basically every byte
475.
▲
by
pierrec
10y ago
Weird. In Ladakh, I ate bags of apricot kernels (which are grown locally and sold in large quantities without any kind of warning, shelled and ready to eat. It hasn't caused any mass poisonings AFAIK). Judging roughly from the picture,
476.
▲
by
pierrec
10y ago
The US today isn't in much of a different boat. Granted, a human being a mushroom is quite an extreme one, but I'm sure a well-crafted documentary reaching a wide enough audience could still have a deplorable impact. Here's a
477.
▲
by
pierrec
10y ago
There are tools that do this, though of course they analyze the source, not the binaries like what's done here (binaries can legitimately contain lots of duplicate code, introduced by the compiler for optimization). https://
478.
▲
by
pierrec
10y ago
OK, makes sense. The headers at the end of the file are only required if you want to extract specific files out of order. Love the clarity of that spec, btw.
479.
▲
by
pierrec
10y ago
I wonder how their API degrades when the stream doesn't support seeking (as is often the case over HTTP, or like in gp's example of using pipes). The whole doc you linked to, and the feature of on-the-fly decompression, seems to a
480.
▲
by
pierrec
10y ago
To clarify: if it's HTTP, you can request the end of file "header" first if the server supports the `Range` HTTP header (many servers support this when it's appropriate). So it's not doable by simple bash-style pipi
More ›