solnic.dev
🛠️ Building @justcrosspost.app & @repobot.app
👨💻 Elixir/Ruby SDK dev at @sentry.io
💜 Creator of Elixir Drops 💦
❤️ Creator of rom-rb and many dry-rb gems and former @Hanamirb.org core team member
#ElixirLang #RubyLang #OpenSource #BuildInPublic
1,209 posts
2,670 followers
457 following
Regular Contributor
Active Commenter
comment in response to
post
Thank you, I hope so hah :)
comment in response to
post
Thanks Lars :) and yeah, it comes and goes, depends so much on what's going on around you
comment in response to
post
If you struggle with stuff, feel free to reach out. We gotta be here for each other 💜
As always: "life's swinging hard but I'm swinging harder" 💪🏻
comment in response to
post
Despite all of that, I have some cool things cooking that I'll be sharing with y'all during the summer and I'm quite psyched about this. My goal is to force myself to do less but more meaningful work, but man it's hard when your brain always wants to do everything, all the time, all at once 🤣
👇🏻
comment in response to
post
My past experience with stimulants has been so-so and I know that good sleep + a lot of sport helps but I can't have it now, so drugs it is again. I'm worried that it may not work well for the anxiety/panic though but who knows? It's always a trial-and-error with these type of drugs.
👇🏻
comment in response to
post
I'll see what the doc has to say but I hope to try stimulants with quick-release with an effect that lasts only for few hours and I'd be taking them when needed, definitely not every day.
👇🏻
comment in response to
post
I'm in therapy again because of this and tomorrow I've got an appointment with a psychiatrist for the first time since 2021. I'll most likely end up on ADHD meds again because normal life chores are hard if not impossible these days.
👇🏻
comment in response to
post
Yeah in general the longer the convo, the worse results you get. Good tools warn you about that and remind you that maybe you should start a new convo. That's why the idea of "memories" is useful.
comment in response to
post
Russia is a terrible country, oh wait
comment in response to
post
I reckon this is guesswork at this point. In some cases it is like you say in others it is absolutely not. We will see the actual impact in years from now. I should mention that personally I root for AI coding myself and use it everyday 🙃 I just try to be as mindful as possible when using it.
comment in response to
post
You missed the "wasteful part". I don't have the numbers but if you consider how much longer things may take it could be significant when global scale is considered.
comment in response to
post
Remote agents are super wasteful though. The number of silly iterations that an agent can go through can be insanely large vs working with the agent realtime to stop it when it goes off the track. I agree that it is impressive and it does get better very fast but it is not "production ready" IMO.
comment in response to
post
That's how it felt for me too but I lacked confidence to call it out 🙃
comment in response to
post
Ah good to know thanks!
comment in response to
post
Yes but we forget about it 🙃 I'm just saying logging inspected structs is a bad idea for at least two reasons
comment in response to
post
It's also not a good idea for security concerns as you may easily leak sensitive data this way.
comment in response to
post
SV would be so wild if it was during AI hype 🤣
comment in response to
post
There are entire studies about that. Keywords: "leaders sociopathy dark triad" 🙃
comment in response to
post
It's literally people like you who make me still support X as this has been my initial motivation, so thanks for reminding me of that 🙂
comment in response to
post
I thought that you can use humor in good faith, that's how I see this 🙂 I also have trouble understanding how this is a straw-man to the point where I had to go and read its definition and I still don't see it in the fly article 🙈
comment in response to
post
I don't see how Glyph's article is related though. The fly one is specific to programming with LLMs and calls out people who don't really know how to leverage this tech but claim it sucks. At least that's my understanding.
comment in response to
post
I stopped posting on X months ago 😢 Lots of people are still there though but I'm making new friends over here and mastodon so that's good 🙂
BTW I'm not yet sure if JCP will continue supporting X in the future. It doesn't feel good to give $$$ to Musk.
comment in response to
post
Same but I don't have time to switch. I guess I just have to suffer 🙃 My biggest frustration is sync between devices. Background app refresh is a joke. Critical things like cal events don't sync reliably on the watch. Gah.
comment in response to
post
Not yet BUT I'm planning some big changes, which includes adding an API 🙂
comment in response to
post
Oh no 😥 Thanks for the info. I now remember reading about it at some point actually. Gah.
comment in response to
post
Thanks for ruining my Wednesday morning 🤣 Why do you think FF is doomed?
comment in response to
post
comment in response to
post
I actually thought it was about the one from fly.io blog 🙃
comment in response to
post
😶🙊
comment in response to
post
Yes but only because I stumbled upon it accidentally over here 🙃
comment in response to
post
comment in response to
post
Yeah man things change so fast it's kinda crazy. That's why I stick to Claude because bothering with all the models and editors would be a full time job 🤣
comment in response to
post
Record a voice note, transcribe and send to your note-taking app
comment in response to
post
Pointing to official docs and sample generated apps should help a lot. I tried that exact upgrade with sonnet 3.7 and it only worked when I gave it specific examples and docs. Claude 4 is much better btw :)
comment in response to
post
If you try to follow what's happening in the AI world on socials, you probably know what I mean.
Basically every day X kills Y because they updated their Z. Alternatively X destroys Y. Every day.
A lot of death and destruction 🤣 ☠️