dastbe

Google at least at the time had a vested interest in making external blaze a thing AND the options in open source were just "not as good". While buck2 is IMO technically better, I don't think there's a compelling reason to switch and I just don't see Meta having a long-term vested interest in keeping it open and growing it.

As a person who probably has one of the best vantage points on this, how was Apple to get apfs out so quickly compared to filesystems in Linux like bcachefs?

asking since ben does take a look here...

will revamped litestream have a solution for ACKing only when transactions have durably committed to storage?

s3 buckets being public by default was stopped 2 or 3 years ago: https://aws.amazon.com/about-aws/whats-new/2022/12/amazon-s3...

longer if using the console

Did you mean L5 and above got more work done with WFH? Since the next sentence implies that it was the fresh engineers who were most impacted by WFH.

Sure, but that's not what the parent said.

?

s3: https://aws.amazon.com/pm/serv-s3

s3 express: https://aws.amazon.com/s3/storage-classes/express-one-zone/

cross-region replication: https://docs.aws.amazon.com/AmazonS3/latest/userguide/replic...

720 degrees is spinning in circles and ending up pointed right back where you started.

there's also a significant amount of automation in place these days to sass you in the right direction, i.e.

* focusing on resources and operations on resources

* using consistent and expected naming schemes, pluralization, etc.

it also helps that the sdks and clis are very raw wrappers around this, such that if you know what it looks like in the sdk then it will look similar in the cli.

with the open sourcing of smithy and the raw s3 specs, open source should be able to build clients in just about any language they care about while keeping relatively high fidelity with implementations s3 releases.

all the other competitors are unserious and/or treating it as a money siphoning scheme. in particular, the public contractors siphoning money from the government and blue origin siphoning money from bezos.

It would be surprising if spacex wasn't just awful at paying things on time, as there's no reason they should have actual financial difficulties.

Eggs are definitely a grey area for vegetarians, to the point where vegetarians will describe ovo-lacto-vegetarians where they are ok with eggs (and animal milk) whereas others aren't.

How much of this was how bazel works vs. starlark itself?

I find the starlark language is very simple (though inconsistent between the various implementations in bazel, go, and rust) but it takes a bit to understand how the magic between defining rules and implementations works in bazel. and TBH, that is also one place I've really needed auto-completion/static typing in starlark to understand what I can/cannot do.

in mba brain, but hosting providers want as much utilization because it means people are going to use more over time.

Part of the problem with pancreatic cancer is detection. Most people (like my dad) are only catching it when it is stage 3 or stage 4 and you're already way on the wrong side of the odds. If you catch pancreatic cancer as stage 1 or stage 2 in most cases the odds are basically 50/50.

whiles taylor swift has a natural monopoly on taylor swift, you can always have more venues. and each of those venues could compete on price/margin to court taylor swift and all of the other artists. and for venues that want to compete, there could exist alternative platforms that also compete on price/features. this all could lower prices AND put more money in the artist and venues pockets.

ticketmaster ensures this doesn’t happen by controlling the market, so venues that don’t use them get locked out from artists and artists who don’t play ticketmaster only venues get locked out from too much of their audience.

it's (1-p)^24^10, where p is drive failure rate per year (assuming it doesn't go up over time). so at 1% that's about 9% or a 1/10 chance of this result. Not exactly great, but not impossible.

the backblaze rates are all over the place, but it does appear they have drives that this rate or lower: https://www.backblaze.com/blog/backblaze-drive-stats-for-q2-...

unless something has changed in the past year, fargate still runs each task in a single use ec2 vm with no further isolation around containers in a task.

my issue with grpc is that looking at https://github.com/grpc/proposal all of the proposals are XDS-related, and its not clear to me why XDS is being so heavily pushed outside of potential GCP interests re: traffic director. Is there really nothing else to work on here?

> If you revoke a priv, things stop working immediately.

for a human-scaled definition of immediately, perhaps, but IAM updates are eventually consistent across the globe.

If you truly need that kind of perf (and at Amazon, we had plenty of dev desktops running on ebs without that kind of performance) then you should really opt for an instance type with local storage.

> Why would anyone contribute to a commercial product unless they're getting paid to do so.

Because they'd be getting paid to do it for their company? I know of a few customers who, if they could, would have their employees contribute minor features to AWS services to solve issues.

What you're saying sounds the same as "The output of one service is the input to another, so the label isn't really clarifying anything".

There's plenty of value in breaking apart all of the interactions between various parts of the system and looking at their outputs and how they intend to change those outputs through which inputs.

goals.

taking a different example, you might have a goal to lower outage minutes by 50%, so your output metric is “outage minutes”. in order to change the output, you need to identify what projects/mechanisms you can use to change the output. More often i’ve seen these be referred to as input goals (increase test coverage, adopt feature flagging, etc.) but you could always associate a metric to them.

the US already pays for 2+ aws region for cia/dod. why not pay for a region that is only available to researchers?

is there any active work on an oss storage backend? any blockers in the jujutsu codebase to starting one?

at least in grpc-java, we've ran into a number of issues with breaking changes that are incredibly subtle. For example, they changed the behavior of the DNS NameResolver to not periodically re-resolve endpoints. This meant that if your Loadbalancer didn't explicitly force re-resolution at some point (which isn't required for any other NameResolver implementation) then you could end up with stuck clients.

? both of these things are available in openjdk distributions

the switch to dot aws is meant to long-term decouple aws api endpoints from a domain that is also used for a bunch of other stuff, like marketing websites and docs. of course, like all aws things they have no consistency over adoption.

as for using a different hostname for dual stack, the reasoning is that they cannot guarantee that all clients of a given service will handle dual stack dns correctly (ex NAT filtering, IAM policies, &c). so they opt to give clients explicit control over migrating to dual stack.

dude pretty clearly got his cheeks clapped by everyone around him, presumably including the board/vcs who invest in carta.