Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
Vogtinator
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
24 ms
·
271.
▲
by
Vogtinator
9y ago
Doesn't this still allow to read memory from the hypervisor shared between the PV VMs in a HVM container?
272.
▲
by
Vogtinator
9y ago
This is not true, the NT kernel itself is the hypervisor.
273.
▲
by
Vogtinator
9y ago
Which is really not a bad vulnerability as the data is still encrypted. Physical access means root anyway.
274.
▲
by
Vogtinator
9y ago
If by "most" you mean those with just a single maintainer who works on it in his free time, you're probably right. For KDE, not so much.
275.
▲
by
Vogtinator
9y ago
That doesn't have anything to do with experience, that's just a fact. Advertisers pay ABP to get on the list.
276.
▲
by
Vogtinator
9y ago
4. Is absolutely wrong in every aspect. Most "innovation" was bought tech (primesense, for instance). The ARMv8 ISA was finalized ages before the chips got designed. Middle of 2011 IIRC.
277.
▲
by
Vogtinator
9y ago
In the first part, the author makes several mistakes including confusing O_DIRECT with DMA. So take this with a grain of salt.
278.
▲
by
Vogtinator
9y ago
Software emulation is fine (otherwise QEMU would be pretty dead), it's just hardware accelerated emulation and actual HW implementations of the ISA that Intel does not like.
279.
▲
by
Vogtinator
9y ago
KDE Plasma has support for redshift as well.
280.
▲
by
Vogtinator
9y ago
And also the worst, from my experience. The good autocompletion isn't made up by the incredible slowness and latency as you type.
281.
▲
by
Vogtinator
9y ago
With compositing enabled that doesn't really matter.
282.
▲
by
Vogtinator
9y ago
That's a different bug then, the one I'm referring to was introduced in 5.9...
283.
▲
by
Vogtinator
9y ago
Testing shows that it works better than Qt 5.8.0, but that's rather easy. Some regressions introduced by 5.8 (some wayland related) aren't fixed yet and Qt WebEngine does not build on AArch64. There is a bug that breaks applicatio
284.
▲
by
Vogtinator
10y ago
That the smart branches optimization requires a workaround in the IR codegen concerns me a bit. It makes sense in a HW context, like MIPS branch delay slots, but not really here. There could be a flag in the branch instruction that disables
285.
▲
by
Vogtinator
10y ago
Linux had that as well until it was castrated a while ago. It was called "exec_domains".
286.
▲
by
Vogtinator
10y ago
Likely everything except GNOME 3.
287.
▲
by
Vogtinator
10y ago
G++ is able to optimize virtual calls to non-virtual calls in most cases (speculative devirtualization) by adding quick checks into the code (like if(obj->vtable == Class::vtable) Class:virtualMethod(obj);) which the JVM would do as well
288.
▲
by
Vogtinator
10y ago
There is a small type difference in C, sizeof() of those two types is different.
289.
▲
by
Vogtinator
10y ago
You can just develop alternative tools and pipe JSON. Text and data in one. Pipes can transfer arbitrary data, so it's just the tools that you don't like, not the underlying mechanism.
290.
▲
Raspberry Pi Gets Its Own Version of SUSE Linux Enterprise Server
(suse.com)
1 points
by
Vogtinator
10y ago
|
0 comments