_wmd

19 ms·

181.
by _wmd 8y ago
Just ask any current/former employee, it's practically folklore by now (i.e. knowledge older than 6 months)
182.
by _wmd 8y ago
You clearly haven't spent much time with App Engine
183.
by _wmd 8y ago
Just another case of their idiotic culture leaking into their products. An early "design" decision led to all their production kit being set to US/Pacific, triggering frequent DST bugs, and last I heard (many years ago) it wa
184.
by _wmd 8y ago
sendmsg isn't a synchronous operation, and there is no lock-step operation of sender and receiver supplying and retrieving a result. Sure you can make something QNX-like, but parent's point was much more general than that: most OS
185.
by _wmd 8y ago
OP wants code to legislate trust problems within a small, closed group, however no amount of code can ever solve trust problems within a small, closed group. No amount of code thrown at people mismanaging their relationship with other peopl
186.
by _wmd 8y ago
Technology cannot solve social problems
187.
by _wmd 8y ago
It seems you've mastered some dark art unknown to an entire industry, you should totally write a book Meanwhile I'll rely on the smelly old debugger to dig me out of threading weirdness, memory corruption, resource leakages, other
188.
by _wmd 8y ago
This reads more like a no-hire report for @ayande than whoever they were interviewing. Turning a bunch of confidential interview notes into an article basically about yourself.. whoa. The only meaningful tidbit is buried at the end: of cour
189.
by _wmd 8y ago
Ignoring the lunacy of wasting decades of training to prove (I hope, to everyone) the blindingly obvious, this would potentially be a real fun experiment to design.
190.
Did You Hear We Got Osama? (2012)(web.archive.org)
3 points by _wmd 8y ago|1 comments
191.
by _wmd 8y ago
Python's allocator makes use of the fact the GIL must be held to avoid any locking in the normal case for objects under 512 bytes. The introduction of pymalloc was a huge perf win at the time. For larger allocations, it's not so c
192.
by _wmd 8y ago
Confirmed uBlock causing it here too
193.
by _wmd 8y ago
It's a referral-only group where you get to play with and complain about their tech before everyone else does
194.
by _wmd 8y ago
The Washington Post link is subscription-only at least for users in Europe, and the video itself is hosted on YouTube. It should probably be reverted to the original link
195.
by _wmd 8y ago
> you can do an awful lot of work if you have an awful lot of work. This is a great way to put the trouble with lots of complex tooling, not just oversized computers. Thanks!
196.
by _wmd 8y ago
300 comments in one hour will definitely kill it. HN penalizes controversy, which it uses comment count as a proxy for. It works well most of the time
197.
by _wmd 8y ago
It's valid concern, but I'm not so sure in this case. Spear phishing is a skilled art, and requires relatively significant knowledge of the target and their domain. Sure the rest is a essentially a stackoverflow post away, but it
198.
by _wmd 8y ago
They've undercut Amazon EFS by 33% for the budget option ($.20/gb vs. $.30/gb). Hope this kicks off a little pricing war - shared filesystem is a hugely useful solution for migrating old apps, would be nice if it were cheap.
199.
by _wmd 8y ago
this video is a 100ft overview, original link is about SQL internals
200.
by _wmd 8y ago
This is very welcome news to me, and I've seen it nowhere else. Do you have a canonical source for it? How about the RAM?
201.
by _wmd 8y ago
A modern Xeon is 20 lanes of PCIv3, or 273MB/sec per drive just to get data into RAM, assuming the bus was exclusively dedicated to disk, which means at full utilization you get about 69k 4k IOPs per unit rather than the rated 500k. Th
202.
by _wmd 8y ago
Note that 2U box needs something like 250gbit internal bus just for disk IO, assuming you wanted the full throughput of every drive.. there is a huge trade off in terms of processing capability and networking bandwidth putting so much disk
203.
by _wmd 8y ago
Heh. You can barely move your mouse across the screen without some layer in the architecture executing some SIMD instruction. It's everywhere, and it's going nowhere. Everything from counting the length of a string to rasterizing
204.
by _wmd 8y ago
Arbitrary package attack - fixed >4 years ago, requires explicit config to enable downgrade, but some distros screw that up Replay and Freeze attacks -- Ouch! sounds like every repo should set an aggressive Valid-Until (24 hours?). This
205.
by _wmd 8y ago
Everything here, but with an important tweak to step 1: it's impossible to know what needs fixed until you've actually gone through the motions of trying things as they are today. It's easy to come in pushing all your 'b
206.
by _wmd 8y ago
The parent context asked what code could have 40 cores spinning, I picked the first example I could think of. You said it wasn't a spinlock, it was. Sure there is a followup lock that sleeps for the empty-queue case, but that doesn&#x2
207.
by _wmd 8y ago
You're among the 1% of engineers who both care enough to know and actually know about these kinds of facilities. For the rest of them, whatever they got at school or off Stack Overflow is as far as it goes, unless some business priorit
208.
by _wmd 8y ago
The accept call on Linux: - invokes the accept() function of the socket family: https://github.com/torvalds/linux/blob/be779f03d563981c65cc7... - which invokes the accept() function of the protocol: https:&#
209.
by _wmd 8y ago
You could easily find examples of this - for example in any textbook threaded socket server design (thread pool contending on accept())
210.
by _wmd 8y ago
The stuff he quoted seems to make a strong case that the behaviour they implement was the intended behaviour prior to Skylake, why else specifically call it out in the newer manuals? Would be interested to know why you think the code was al

More ›