HN user
This exact problem had me worried when I heard about OpenBSD's arc4random, and how they've been promoting a pervasive use of it. I haven't taken the time to look at how they solve the problem of contention, while still maximizing the unpredictability of the random number generator's state by using it.
That would be interesting, however after launching adb shell:
shell@mako:/ $ grep vfat /proc/filesystems
vfat
shell@mako:/ $ grep vfat /proc/mounts
/dev/block/platform/msm_sdcc.1/by-name/modem /firmware vfat ro,relatime,uid=1000,gid=1000,fmask=0337,dmask=0227,codepage=cp437,iocharset=iso8859-1,shortname=mixed,errors=remount-ro 0 0
Just the other day, I noticed that on my kde desktop, wireshark was the only program using gtk3, whereas gtk2 is used by a bunch of cross platform programs. I'm not too sure what that means though.
Putting a dollar amount on anything signals value perception. 12.50$ is a lot worse than a warm welcome, or other free rewards like public acknowledgement, because it says Yahoo really couldn't care less about finding such bugs.
I notice the ideas that it is not the number of contributors that matter, but the number of sufficiently skilled ones, and the that popularity impedes change. I can't help draw a parallel with the advice that you should listen to your most valuable customers, and potential customers, and that the rest of your users will expect free stuff, and complain loudly when you pivot.
Also, cache invalidation can actually be cache replacement. Otherwise, under very high load, you risk cache stampede.
The unsafe language argument of the first paragraph doesn't hold. You can design you API such that handles are copied by value, and opaquely contain pointer. Usually the handle IS the pointer, but it doesn't have to be this way. When you do this, you are able to perform exactly the same kinds of memory operations that a virtual machine would perform.
I too wondered why he would ask Apple rather than the DOJ. If he did the latter, the answer would be that they don't know the details of the implementation, and can't say. Apple is the party which knows right now.
So what I would like, is that after Apple comes out with the details, the question gets asked to the government. That would ensure that the government does behave according to the intent of Congress.
Stanford has another somewhat related project: http://dune.scs.stanford.edu
In their case, the application runs both as a process in the host and as a guest. It gives the application access to traditional OS APIs, and allows use of processor extensions to directly access virtualized hardware. The benefits of doing this include ability for an application to do low level custom IPC, to use the page tables for garbage collection, to trace the use of system calls much more efficiently, to hook into page faults, and so on. Very cool stuff.
In that case, your IO problem is easy, because it's small with regard to CPU time. You can get away with putting the whole data in sql databases, and/or making multiple copies of your data. Then you can use as many workers as you want, with usually simple partitioning logic.
In your specific case, you could just just grab all the classes in a big list [1], and pick at random from them.
You don't need to do much to figure out the url from the names, so it's really only the list of names you need.
[1] http://docs.oracle.com/javase/7/docs/api/allclasses-frame.ht...
You have to give credit to languages like java or C# which provide the programming interface wich does the right thing. Those who use lower level languages because they want better performance should reconsider unless they have the required know-how. It baffles me that someone would write C++, and mindlessly concatenate string.
It's a shame they have actually converted a real system, and they rely on the language benchmark to make an argument about performance.
That said, not having run into a problem is a worthwhile feedback.
Avoiding double journaling Looks is another case of the End-to-End principle. Nice catch.
On the other hand, I would have expected a performance study to include a power consumption discussion, especially on mobile.
While playing these kinds of games on security sensitive services might not seem like a good idea, it lets someone take ownership of its development, and gain experience with its codebase.
I too wish for such deals. I'd particularly like that for textbooks, where I expect to want to refer to them later on, and possibly when I'm far from my copy. I can't think of a good way to convince the editor that I haven't resold the paper version though. That would end up costing them a lot.
Deterence is increased both by the severity of the punishment, but also by the certainty of receiving the punishment. I think most major crimes are committed by people who have comitted minor crimes. If getting caught for minor things becomes more likely, it may be less tempting to go down that path. There's a large body of work trying to evaluate the effects of both causes. Arguably, the US has gone way down in the realm of diminishing returns in terms of severity, especially when it come to drug laws. Of course this doesn't take away any of the potential downsides in terms of civil liberties of any approach that raises the certainty of getting punishment.