kmarc

I can highly recommend OpenCode Go.

I use it from pi.dev as well through the OpenCode Go $10 subscription ($5 first month).

Used more than 20M tokens at a cost of ~$20 (up to $60 is included in the $5 plan) Out of which deepseek pro had ~200 messages which is around 1.5M tokens (10+M cached)

After 18y of thinkpads, this year I bouth a Lenovo yoga for... Cheap (1000€).

32G RAM, nvme 1TB, core ultra 258V.

Looking at the prices now... Wow, was I lucky.

Tried some of the 7b models locally, more than usable, around 30token/sec, not with the NPU, but using the ARC integrated GPU.

I am a noob for this, but I guess it's time to experiment more with this local setup

Does a nazi salut make him one?

Regardless, the technology itself is amazing, and eventually should (will) be common infrastructure

https://github.com/CyberShadow/aconfmgr

Aconfmgr is a good solution (practical trade-off for me). When it feel like "my OS must be dirty at this point", then I spend a couple minutes with this and I'm back to the clean state.

I don't do this more then twice a year, so if you insist on an always clean system, YMMV

Stalled...? I even had to look up the name using "pgp social network xlm" as search terms.

App still working tho.

I use Keybase's encrypted git repo the same way (to sync "private" dotfiles across laptops / remote workstations)

I visited the local waste processing plant to learn about the basics in general and have some insights on what could I do for better recycling.

Clothing is the worst offender here. Impossible to recycle. The quality would not be good enough and the price is too high. It is a huge waste of resources.

Upcycling (e.g. repurposing) works with some materials, though, however, with very low efficiency.

Europe is moving towards reducing fast fashion therefore. This is a good thing in general, but I don't see this change in the US / Asia. And very slowly also in Europe.

Wait, isn't that what Windows 3.1/95 did with the "What's this" button?

This resonates with me a lot.

I do some stuff with gemini flash and Aider, but mostly because I want to avoid locking myself into a walled garden of models, UIs and company

Do you know how it compares to LibreTube[1]? For the SponsorBlock integration; it works well for me, but I kinda miss the newpipe interface.

[1] https://github.com/libre-tube/LibreTube

I used both my thinkpads for 8+ years, generating $$$ for me. Build quality matters a lot when it comes to longevity, thinkpads were great at it decades ago. I would expect the same from MacBooks nowadays (I also had an M1 Pro Max from a former employer).

Funnily, this is how I work, exactly.

I want to believe that by getting back on track from nothing to where I left off already helps a lot to understand the problem at hand, and maybe realizing issues that before prevented me of finishing the task.

Haven't used this app for ages, but yesterday I fine-tuned the HDR settings for my newly bought laptop under KDE/plasma, definitely on Linux. That readme might be outdated (or the tool doesn't work on KDE at all)

This was an interesting read for me. I'm mostly aware of the _problem_, however, never wondered how that could be fixed with other designs, I guess he is working on something that implements one of his proposals (river/window/bonfire)

I am one of the hoarders who has saved Inoreader items, a "Later" bookmark folder with (once thought as) interesting stuff in it, obsidian we clips for the ones what are so precious I for sure didn't just want to reference to but actually make a copy of. But it's under control. It doesn't give me anxiety knowing that I "should" go through them, because... I often do.

I'm surprised that the "first" of these layouts only appeared in 2002. I would have sworn I used Akregator since 1999

Hah, I guess this is the same then as in formula 1 (and possibly other motorsports). After the end of a straight (speeding) section, just before the curve they have 150/100/50 (m?) distance indicators. Sans the concrete block at the top. That would obviously shave the driver's head off.

When a driver hits these, they evaporate as dust.

Just finished it, and while I loved the whole plot, the adventurous expeditions away from the base, somehow this one with the waaay too long paragraphs seemed... Unnecessarily boring?

My first Tchaikovsky was children of time and TBH none of the sequels nor his other space operas were as captivating as that one for me.

Yet, I will read this one too. I believe that his ideas and stories are great in books and would never be able to make them into movies. So unique.

Unfortunate; This morning my firefox profile didn't start up, draws the empty window, and spins the CPU. I started my workday with downgrading FF and restoring my previous profile from a backup.

Now this.

FWIF, for this particular problem, the workaround from the linked reddit post works:

Edit 3: This workaround seems to fix it:

    type about:config in the URL bar
    search for network.http.accept-encoding.secure
    replace gzip, deflate, br, zstd with gzip, deflate, zstd

I'm in the same boat.

Apparently a group of devs forked it: https://github.com/dwash96/cecli

Haven't tried yet

I'm 40 and sometimes I feel the same. Should I be worried...

I had to do something similar. Ruby is awful and very immature compared to python, so I "outsourced" the machine learning / LLM interaction to python. The rails service talks to it through grpc / protobuf and it works wonderfully.

Damned is this industry, when even _you_ say you have to show that "remoteness works".

I also measure meetings (counts, lengths, and mostly meeting minutes/outine jotted down by myself) and keep track of other metrics, exactly for this reason. However, I also don't happen to have written best selling books and stuff, so I really must do this, and you really shouldn't have to :-)

Davmail is the shim I used to proxy OWA / exchange ansmd present an IMAP interface towards neomutt. Used this setup with three companies, worked well, and unified the UX instead of the different Gmail and Outlook web slowness.

I guess the relevant point I missed to emphasize is that I would have hoped that the situation is becoming better with the next major version, but as others also pointed out, it doesn't.

I feel you. Rubocop is certainly rather a bikeshedding tool than a linter at this point.

The forced if/else transformations drive me nuts.

I'm tasked to amend a project written in ruby. With a python background (and some nice pydantic, type annotated, etc "strong" code bases behind my back), every day I spend with ruby is a minefield, a nightmare.

I hoped that ruby4 maybe implements stuff that python has, like type annotations or making the damn parens mandatory, but no. Not surprised that python has ten times more developers according to stackoverflow's survey... I can't possibly imagine a collaborative project where other people also have to work on the same code base, and not having any clue what a symbol under the cursor might be.

No type hints. No mandatory requires. No parens, so never know if something is a method or, callable, or variable. Basically IRB is a must for development, because in the editor, I'm blind.

And the ecosystem is just sad. Swagger-rails libraries out there are rookie jokes compared to what python has. At least there is decent GRPC / protobuf integration, so all new services I am writing can be in python. Or any sane language.

Yesterday I decided to finally write my makefiles to "mirror" (make available offline) the docs of the libraries I'm using. doc2dash for sphinx-enabled projects, and then using dash / zeal.

Then I was like... "when did I last time fly for 10+ hours and wanted to do programming, etc, so that I need offline docs?" So I gave up.

Today I can't browse the libs' docs quickly, so I'm resuming the work on my local mirroring :-)

Same, with an x1 gen5, upgraded NVME to 1TB

This boy is 8 year old today (bought in 2017 November) and still delivers me the €€€ at $consultingjob

Happened in the Netherlands: https://en.wikipedia.org/wiki/El_Al_Flight_1862

40+ people died in that one, it's a miracle it wasn't more.

I caught myself that most of my LLM usage is like this:

ask a loaded, "filter question" I more or less know the answer for, and mostly skip the prose and get to the links to its sources.

I got an iPad many years ago from my employer. I literally only use it for libby :-)