Added basic link test cases and a cp(2) example using linked read and writes to liburing. Needs a bit of love (short reads, errors), consider it a tech demo.
https://t.co/I31tBWMLdo
Learn about QUIC, straight from the experts. Join @mnot, Jana Iyengar, and @kazuho for a lively lunch discussion about QUIC on May 21 in London. RSVP here: https://t.co/xNhtU8RXcS
Six Stages of Debugging
Step 1 : That can’t happen.
Step 2 : That doesn’t happen on my machine.
Step 3 : That shouldn’t happen.
Step 4 : Why does that happen?
Step 5 : Oh, I see.
Step 6 : how did that ever work?
@FiloSottile mprotect(PROT_NONE) would work, it does have the disadvantage of creating a new vma (continuous mappings need to be split if they have different protections)
These views are based on the strong belief that people avoid strong static typing and structured environments not because they are unenlightened, but because dynamic types and the like are way better for prototyping. (I, personally, love to program in Python like it's OCaml.)
Very impressive improvements to IOPS and latency reduction coming to @Ceph by the use of the io_uring interface which was added by @axboe to Linux kernel 5.
https://t.co/xwaFTHxuYq
@pcwalton@axelgneiting > * I deliberately chose not to comment it. You should have at least
> * as much fun trying to understand it, as I had to write it :-).
wat
One visit to a website, and unbeknownst to you your browser can be enlisted in a botnet and remain under control even when you leave the site... https://t.co/BVR1p9ZDGR
@jedisct1@bagder It's unclear how the proposed solution (further down): "The simple solution [..] is to introduce a new archive mime-type that can transfer chunks of the website at once, including several HTML pages, images, css" would be simpler than the better framing in HTTP/2
Maybe this is trivial, but the purpose of DoH/ESNI is largely defeatet if OCSP without stapling is used. (OCSP is plain HTTP and if you look at browser traffic these days it makes up a large amount of the remaining HTTP traffic.)
Finally, some friends and I are ready to soft-announce the startup we've been working on: https://t.co/u2CdqSZn2W. We're working on a thesis that most modern tech architecture is "too scalable," and that slows down development of features that actually matter.