awinter-py

homebrew the package manager that is literally built on git yet still doesn't have version pinning or versioning for most packages so that you have to host your own copies of public packages to use the old one?

or a different homebrew

1 amazing work, but 2, zooming out:

OSS at some point forgot that the web + hosted UX is great for saas companies because they can more easily charge rent. But the web isn't great for building input UX (it's all bloat + jank and you have to completely reimplement the input elements to do anything useful, on top of shifting sand).

moving UX back to desktop thin clients would be leapfrog innovation over the saas companies. I include things like MS word as a 'thin client'. MS word 2003 was more responsive than any web interface for just typing. slack has like 2 features and is still somehow sinking into the ocean unless you have a supercomputer because of electron.

the data can be shared or backed up somehow sure; that's just file hosting with a collaboration layer maybe. but desktop UX would be a superpower for any tool that needs powerful input

with larger corp users there is often an MSA that overrides the TOS; no idea if that's the case here, but give hospital GC benefit of the doubt re due diligence

also would be interesting if crowdstrike was installed by a reseller who specializes in healthcare / airlines, not the hospital or airline itself

waivers not always enforced, per [1]

> Courts will hold overbroad liability waivers unenforceable on public policy grounds. ... injurers routinely ignore these holdings and persist in requiring would-be plaintiffs to sign such unenforceable waivers anyway

at least one federal case in florida[2] saying advertising a product as safe doesn't defeat the EULA, but it's florida

1. https://wp0.vanderbilt.edu/lawreview/wp-content/uploads/site...

2. https://casetext.com/case/justtech-llc-v-kaseya-us-llc

welp that was fast

> Transition diagrams for state machines, Swimlane diagrams for request / response protocols.

100%. especially because these things also form the basis for static analysis; if your state / protocol diagrams let you run something like TLA you are doing really good

acemoglu development econ paper on this worth a read

follows postcolonial trajectories of coastal countries used for ports + trade vs inland countries used for resource extraction labor

telegraph needs copper. it just helped find some

fairly sure the Lean Startup book is based on intuit, which is funny given that rather than a product company, they are a lobbying organization so distasteful that even a country generally tolerant of distasteful lobbying finally succeeded in passing them like a kidney stone

their uv tool is incredible -- it is substantially faster than vanilla pip even on a fast laptop, and they were smart to adopt pip-tools commands for pinning

will not go back to pipenv / poetry unless uv does something truly awful to me

I feel more meh about rye, except as a way to use uv for package management

and I don't understand why astral is a company

I may be late to this discussion, but what has changed in MV3 that shifts the balance of power to publishers? declarative WebRequest?

so quick bright things come to confusion

the old system was fine

related: AT&T / fairshake arbitration https://www.nytimes.com/2020/04/06/business/arbitration-over...

fairshake realized that AT&T TOS had some protections against suits + mass claims, but did allow individual arbitration; and AT&T pays an arbitration fee for every case that is filed

looks like there is still a pcb within the enclosure?

not sure exactly how e-phoretic screens are driven, but I wonder if you could remove the controller entirely? have just a loose eink screen with a zif that you pass around

like this thing is 0.25mm thick https://www.adafruit.com/product/4262#technical-details

just came here to upvote the alphago / MCTS comments

yes this

csv file with emails + python thing to send them using postmark

it took an hour every time I had to do anything, and I ended up with a collection of python scripts that solved random problems that were already solved somewhere else, which grew and grew

I tried a few programmable shared inbox tools and they were all clunky and meh. tried lighter things like supabase + some golang list manager, also meh.

postmark has a CRM they acquired, iirc it was too expensive

zoho seems reasonable on price, but did not try it

but to be clear, even if I paid up for something dumb like hubspot, my life would have been simpler and better

zapier is a good place to shop for these

no comment on this product specifically, but large takeaway from hitting a point where my side thing had actual users + communication -- CRM is critical

if you asked me 24 months ago I'd tell you CRM was a BS category, now I think of it as a backend component on par with the DB

arduino is the other name in this space + has been making interesting moves; their industrial / PLC offering has gotten very serious in the last few years and they are ahead of pi on the cloud front (though open question whether either of them will do well in cloud; their customers are too savvy + nimble for a lock-in play)

arduino is approaching the business bottom up (tiny well-known chips), pi top-down (it's just linux with some GPIO pins)

also have competitors at the margin like nvidia (who could legit own 'iot edge AI' if that ever even matters) + waveshare, the off-brand pi manufacturer who remarkably has way more pi-compatible devices than actual pi

excited for IPO if it means the #1 linux board finally supports suspend-to-ram

'your backend is just files' seems like an idea that is due

in the oss self-host world, countless things that are just file servers (with some sugar) are far more difficult to host than they should be: immich, nextcloud are examples. standard notes is an extreme example, its compose.yml is like the vhs tape from the ring

same in closed in-house backends; at best people store some static files in s3, but we use databases way too much

'your files contain entities' vs 'your files are entities' is a split; this project is the former but I kind of want the latter

also permissions is kind of the hard part as other posts have mentioned here

but hope to see more stuff in this area

at long last we have created the torment nexus

I think it's okay to break some legacy libraries. they can either fix themselves or fade away

better to lose a few libraries to modernization than the whole language community

ughghgh

every proposal for C/C++ modules feels like it's by someone who has not used a working module system and doesn't believe it's truly possible to create

they do 2 billion rides per quarter

what are the 'trillions' here?

^ also that translates to ~1000 transactions per second with some assumptions; have never understood why they care so much about infra scaling

1000 tps is like 1 box

if shitty TOS turns out to be the thing that bends the arc of history from AI to privacy, it would make me so happy

the large kind also do this in a sense

I feel this way about all wrapper code

where someone is like 'okay we need our own internal version of this API'. the reasons vary but are like 'we don't trust people to use the official API correctly' (which, fine, but yours is less standard and worse-documented).

sometimes the answer is 'we need extra functionality' which, fine, but 1) don't wrap the entire API for gods sake, just add 3 functions, and 2) your codebase over time will end up 99% polyfills

point being if you are not using the defaults you are inflicting great misery on whoever inherits your codebase

no argument with the benefit for doctors. I know docs who say they are squeezed because their day is engineered in a way to force them to do paperwork later

just wondering what computer process takes minutes. is it building the linux kernel every time?

> Lall said it takes minutes for AI Scribe to collate the information, allowing her to move on to other patients while the SOAP note is being created in the background.

it takes minutes ...

google's revenue is 300b, the mayan calendar ends in 6 years, do the math + that's about 2 trillion