plonk

17 ms·

181.
by plonk 4y ago
> - The right side bar is "dynamic". Most of the time it's the Explorer showing the directories and files, but when I try to search for something, though, it becomes the Search sidebar. Same for all the potential sidebars
182.
by plonk 4y ago
You need a good machine, but on an M2 SSD with a recent CPU and enough RAM, VSCode takes one second to start, much less to open a new tab. Not sure Sublime Text's speed will still be a big selling point in the future.
183.
by plonk 4y ago
For us the work was mostly in adapting the old codebase to CI/CD and writing/debugging the workflows. There’s almost no additional cost to running self-hosted runners, in fact we wouldn’t be using GitHub at all if they didn’t offe
184.
by plonk 4y ago
It’s great for deployment. I’m still weirded out when the result works so well, when every other Python deployment solution always fails on some common package.
185.
by plonk 4y ago
No source now, but I think I've read that the jet is matter that's far enough from the black hole to escape it, accelerated during its rotation. The wiki page isn't very clear about how the jet is created: https://
186.
by plonk 4y ago
https://math.ucr.edu/home/baez/physics/Relativity/BlackHoles... > So if you, watching from a safe distance, attempt to witness my fall into the hole, you'll see me fall more and more slowly as th
187.
by plonk 4y ago
Hi! Are these positional embeddings literally made by concatenating the patch embedding with a number, then passing that through the next layer, as suggested by the figure under "Images to Patch Embeddings"? It's the most con
188.
by plonk 4y ago
You got downvoted unfairly, asking for sources is totally normal.
189.
by plonk 4y ago
You're right, my comment only applies when bandwidth is the bottleneck. In Python, that web parser could probably do even better with asyncio in a single OS thread.
190.
by plonk 4y ago
If I/O was the bottleneck, parallelizing it won't help, your SSD/network link/database won't get magically faster. If I/O wasn't the bottleneck, I guess you can parallelize reading, but what are you gainin
191.
by plonk 4y ago
You can usually counteract that by sending large enough batches to the processes.
192.
by plonk 4y ago
Not much more than the switch to MapReduce and threads. Actually the interface is exactly the same if you use executors from concurrent.futures.
193.
by plonk 4y ago
Yes I think so.
194.
by plonk 4y ago
Middle school flashbacks. If you have a figure like this, you write this, then this, then "according to Thales' theorem", then that result. No, you didn't write "according to Thales' theorem", the result i
195.
by plonk 4y ago
What I'd have loved to be taught is basic financial skills, as you said. But you need to cram that into an already crammed schedule that overworks children for diminishing returns (I found school miserable) and the math taught in my (f
196.
by plonk 4y ago
Google's core services wouldn't work so reliably if they didn't value optimization and engineering. I don't work there, but I'm pretty sure that the SREs and the developers behind Search and Maps don't get fire
197.
by plonk 4y ago
> An obvious optimization would be to utilize all available CPU cores by using the MapReduce pattern with multiple threads. Nope, the GIL will make that useless. You need to actually implement the tight loops in C/C++ and call that
198.
by plonk 4y ago
I find that people with some advanced math education are just better at reasoning and more successful in general. Teaching some people arithmetic and others the "rest" would be a huge shame. It's doubly damaging in countries
199.
by plonk 4y ago
I think it's good that the author made it as accessible as possible, but had to pause after a few chapters because they took so long to explain the basics (I was always bad at uni-level math but still studied half a math bachelor befor
200.
by plonk 4y ago
The only way to actually run the workflow is to push the code and trigger the workflow somehow. And you can't push code without committing it since this is git. You could write the workflow correctly the first time. Good luck. It&#x2
201.
by plonk 4y ago
I add a workflow like this: - write the main job in a Python/PowerShell/Bash script that runs locally, - write a workflow that sets up the environment (AWS login), installs dependencies (using GitHub's facilities for that) an
202.
by plonk 4y ago
That deprecation was because the Docker shim (compatibility layer) was expensive to maintain. You can still use containerd, which is Docker’s runtime, as a Kubernetes runtime.
203.
by plonk 4y ago
Ever seen a PR that implements something in a GitHub Actions workflow? The history usually looks like: clear cache, fix path, fix variable expansion, fix command, fix command again, fix syntax, […]. The best way IMO is to interactive-rebase
204.
by plonk 4y ago
I wonder why GitHub doesn’t apply their own signature when they rebase a commit with a valid signature from one of their users. They do that when you edit a file through their Web UI.
205.
by plonk 4y ago
Interesting thread by Matthew Green about HME and privacy: https://ioc.exchange/@matthew_d_green/109383332045800279
206.
by plonk 4y ago
And I'd be happy if companies supported Linux because they like free software and not for economic reasons, but the end result is still traction for Linux, so I'm not complaining.
207.
by plonk 4y ago
> When I tried mastodon it completely failed to provide the basis functionality I expect. I think that's because it's technically decentralized, but they've then built a centralized service through social coordination on t
208.
by plonk 4y ago
A lot of migration is fuelled by Musk firing XX% of Twitter's workforce, and several competent people (former SREs at big tech corps, systems engineers) posting threads that predict huge technical issues in the near future. Some also j
209.
by plonk 4y ago
Not that far away now.
210.
by plonk 4y ago
Well I've been stuck like that a few times (still very young though), and changing cities/states/countries has always been the best way to get out of it. There are probably IC positions that don't cause Amazon-style stre

More ›