PWV Portfolio Feed

We need to acknowledge the current status quo where GitHub tags are _the_ way to release Swift packages and it would take quite some time to change that, especially for a registry not officially maintained or recommended by Apple.

As for the comment about the Tuist Registry "mirroring" releases from GitHub, this was the fastest way to provide the most value to the ecosystem and we think it strikes the right balance.

You know you did something interesting when Dave Werver highlights your work in the iOS Dev Weekly newsletter - and even writes up a whole commentary for it 💜

App code migrations are easy, database migrations can be terrifying.

Databases hold your company's entire history.

3 rules:

• Never drop columns in prod without a deprecation period
• Always have a rollback plan
• Test with production-scale data

Sleep better 😴

we’re on a combination of services, this bit goes through cloudfront which is a little less magic afaik

i still think there's real innovation in llms and they'll have a positive impact. but the behavior of a lot of the people in the space - these crawlers ignoring all the norms with robots.txt and companies knowingly training on pirated material - this is more the category of 'people doing bad stuff'

the final step of my llm-related demo today is to write some code to block llm crawlers from hammering the demo website with requests, blowing through my llm bills, this is all good and normal

4/4 Thank you so much for all of your contributions to Zed, Jason! Your work on Longbridge and gpui-component is entirely inspiring to all of us and we are excited to see how everything pans out!

1/4 Jason Lee is a very special contributor who has been helping us out for ~1 year now. With 58 merged PRs and 4 currently open, Jason's PRs have included everything from improving language support, to fixing bugs in GPUI, to rounding out Windows support, and more!

⬇️ new tutorial to check out for some Friday afternoon and weekend projects

That being said, SwiftPM is open source, so anybody is free to poke around and propose optimizations there 👀 But registry will still stay as something needed as there are inefficiencies that are unsolvable with the decentralized source control model.

In other words, the optimization might be possible in _some_ cases – but especially locally, you would probably need a deep clone anyway. We'd be surprised if this wasn't considered when the registry was proposed, so we see it unlikely as being implemented.

Deep clones could probably be skipped if literally every dependency would be pinned to a specific reference/version. However, as soon as you have one dependency that has a range of versions, SwiftPM needs to do a deep clone to be able to resolve the right version.

tax return deadline in a week, still haven't done it, so i guess i'm gonna be super productive on basically everything else

👋 Say hi to Ryan, a Senior Developer Advocate in our #MeetThePrismanauts series

Catch him flying planes ✈️, playing with his kids, or hacking on code and video projects 💻

🇨🇦 Ryan hails from Canada which is much too cold for his liking 🥶

Connect with him at @ryanchenkie!

yay thank you! DMd you :)

my "the typescript team wants to do it but doesn't have enough staff and can't get this work prioritized enough to spend the time on it but the vscode team really seems to have a lot of money and staff and this may indicate something about corporate priorities" sweater is raising other questions

my "this thread in the microsoft typescript repo about why typescript doesn't implement the microsoft language serve protocol to support other editors when the microsoft vscode editor typescript support is good says something about microsoft and market power" tshirt is raising a lot of questions