Sesse__

47 ms·

901.
by Sesse__ 4y ago
> I wish browser vendors could work around this limitation and stick to Sass syntax. FWIW, it is not a limitation, it's a conscious choice. If you want to accept arbitrary selectors in option 3, you'll need unlimited lookahead,
902.
by Sesse__ 4y ago
> One of the ideas of ipv6 was to reduce routing tables That idea was abandoned about 20 years ago. One fairly quickly discovered that hierarchical routing does not work well in the real Internet, where redundancy is done on the IP level
903.
by Sesse__ 4y ago
That's not insane, that's pretty cheap. If you buy IPv4 addresses in bulk (e.g. an entire /18 or so), they're ~$40 a piece these days. When you buy cloud services from AWS or GCE, you're still paying for those IPv4
904.
by Sesse__ 4y ago
> Why are we stuck with nat in 2022 again? Why did we not map the 32 bits of ipv4 to ipv6? Because it does not help. djb's “idea” was to change everybody's setup _first_ and then have a flag day where the entire world were move
905.
by Sesse__ 4y ago
> A cooler feature would be requiring the compiler to prove the addition wouldn’t overflow. Wuffs is specialized enough to do exactly that ( https://github.com/google/wuffs ).
906.
by Sesse__ 4y ago
“use bignum;” and you can.
907.
by Sesse__ 4y ago
> It wouldn't produce an exception, it would not compile. The nice thing is that you can avoid range checking at runtime. You're proposing a language where you cannot increment integers? I don't think that would be a very
908.
by Sesse__ 4y ago
Depends on what you're searching for. For position lookup, there's an index; for something like a given player, it's likely to be a linear scan. Some format documentation: https://web.archive.org/web/2016
909.
by Sesse__ 4y ago
Applying a blur filter to a premultiplied image is (very subtly) _correct_, at least if the goal is to emulate what happens if you used a physical lens to blur the same image. Not only does postmultiplied alpha mess up the correct pixel val
910.
by Sesse__ 4y ago
The CSS Color (Level 4) spec allows you to interpolate in any color space that you can specify colors in, and all of these four are considered useful to specify colors in. (Or at least useful enough to make it into the spec :-) ) This leads
911.
by Sesse__ 4y ago
There has been some movement to handle CSS layout in JavaScript (running in a service worker, so that it cannot have state, but otherwise fully in sync with the layout), as part of the “Houdini” effort: https://www.w3.org/TR
912.
by Sesse__ 4y ago
It probably relates to the tuning you're used to. Piano strings do not have perfectly linear overtones, and thus piano tuning is never 100% equitemperal (I've tried; it sounds quite off). So the tuner has to make concessions acros
913.
by Sesse__ 4y ago
Occam's Razor would suggest that Phoronix' benchmarks are broken.
914.
by Sesse__ 4y ago
It's not low-hanging fruit at all: * The gain depends on low-level details of your storage, can easily be negative, and is not necessarily easy to calculate precisely * It requires adding an extra join, which can be expensive (dependin
915.
by Sesse__ 4y ago
To be clear: MySQL cannot use an index to directly seek past the first OFFSET rows; you still have O(n) behavior for OFFSET n. But you can use a covering index scan for the rows that you skip, which especially for InnoDB is somewhat cheaper
916.
by Sesse__ 4y ago
> If the index had cardinality metadata, it could jump directly to the leaf node containing the first row in the result set. It's difficult to maintain cardinality metadata in the face of MVCC, even for the table as a whole. Page-le
917.
by Sesse__ 4y ago
In general, you cannot push a LIMIT and OFFSET down through a join, since the result is not guaranteed to be the same. (I have no idea what this gem does about that, but maybe it has some sort of 1:1 guarantee through foreign key informatio
918.
by Sesse__ 4y ago
Because BGP doesn't support ASN numbers larger than 16 bits; the 32-bit support is a pretty nasty hack, and for a long time, lots of equipment would not support it.
919.
by Sesse__ 4y ago
> Would Nike be happy if Portugal’s Cristiano Ronaldo suddenly appeared in Adidas shoes? Back when Nike Vaporfly came out and the other shoe manufacturers de facto didn't have an answer yet, you would see e.g. Puma-sponsored athlete
920.
by Sesse__ 4y ago
Some things are amazingly slow, though; look at how long it takes to install software, for instance. That's rarely CPU-bound at all, but somehow is influenced by I/O mismanagement in the various installers.
921.
by Sesse__ 4y ago
I live not far from Valgrindvegen (Valgrind road); I've always wondered whether the developers knew it existed. :-)
922.
by Sesse__ 4y ago
If you only want memory leaks, LSan will do that for you. In general, I tend to use ASan for nearly everything I used Valgrind for back in the day; it's faster and usually more precise (Valgrind cannot reliably detect small overflows b
923.
by Sesse__ 4y ago
B-trees don't work well for CPU caches; 64b lines are typically too small to gain much. (OK, the M1 has 128b lines, but still.) And you still get the branch mispredicts, and a significant increase in code complexity, so hashing is stil
924.
by Sesse__ 4y ago
> Ironically, due to caches, sorting and then using algorithms that rely on order tend to be superior than most hashing implementations This doesn't match my experience at all. C++ trees are not cache-friendly; they're pointer-
925.
by Sesse__ 4y ago
One thing Hex-Rays has that Ghidra doesn't (and cannot) is amazing support. Back when I had a license at work, I could report bugs and literally get a fixed binary back a couple of hours later. They're both amazing, they're b
926.
by Sesse__ 4y ago
If it's double, it probably would be addss xmm0, xmm1 :-) (add eax, ebx would be for 32-bit integers.)
927.
by Sesse__ 4y ago
LuaJIT.
928.
by Sesse__ 4y ago
Because Guido van Rossum just isn't very good with performance, and when others tried to contribute improvements, he started heckling their talk because he thought they were “condescending”: https://lwn.net/Articles&#x2
929.
by Sesse__ 4y ago
You are mixing up IPv6 prefix rotation and IPv6 privacy extensions, and you don't seem to take into account that IPv4 from most ISPs is much worse (typically, you get an IPv4 address from your ISP via DHCP and keep it nearly forever, n
930.
by Sesse__ 4y ago
I had a system that was sped up by 30%+ on Windows by switching from HeapAlloc to jemalloc. Profiling showed that HeapAlloc was largly stuck in a single giant lock. (This was on Windows Server 2016, IIRC.) And that wasn't even that all

More ›