mkesper

32 ms·

1. 301.▲
   by mkesper 3y ago

   Who uses this and can compare against FastAPI?
2. 302.▲
   by mkesper 3y ago

   From the article: VA: This lack of effective leadership from the public sector comes up over and over again in your discussions of American transit, both internally and in their relationships with other actors. They struggle at cooperating
3. 303.▲
   by mkesper 3y ago

   For projects without Logo it would be nice to autogenerate some label.
4. 304.▲
   by mkesper 3y ago

   Probably because our school atom model is too simple. https://www.quantamagazine.org/inside-the-proton-the-most-co...
5. 305.▲
   by mkesper 3y ago

   Please add (something like) that to the Readme.
6. 306.▲
   by mkesper 3y ago

   I let you in on a secret: If you run a docker container on MacOS, you're running it in a VM. On Linux, you don't (at least not if you don't install the braindead Docker Desktop) https://docs.docker.com/desktop
7. 307.▲
   by mkesper 3y ago

   But the migration stuff is a horrible joke. No way to just rollback a broken migration. https://www.prisma.io/docs/orm/prisma-migrate/workflows/gene...
8. 308.▲
   by mkesper 3y ago

   Please evaluate the license of the extension package carefully before using!
9. 309.▲
   by mkesper 3y ago

   Or maybe this: https://www.youtube.com/watch?v=uxMmaygEYP8
10. 310.▲
    by mkesper 3y ago

    You could use mypy in pre-commit so code is always checked. Sad there's no per-file switch to tell the interpreter 'this has to be checked'.
11. 311.▲
    by mkesper 3y ago

    I bet your AMD CPU has an internal GPU, too. That's what you can use with the unified memory.
12. 312.▲
    by mkesper 3y ago

    Well, yes, you need to install ROCm and depdendencies. Have a look at https://rocm.docs.amd.com/projects/install-on-linux/en/lates... Debian trixie (not yet released) has most dependencies as packages. Or you
13. 313.▲
    by mkesper 3y ago

    Or a Linux machine with a Ryzen using the internal GPU and the unified RAM (scroll down at llama.cpp and look for ROCm).
14. 314.▲
    by mkesper 3y ago

    Use doomemacs for a start. It really optimizes startup time and offers vast included modules as well as great package management. https://github.com/doomemacs/doomemacs/blob/master/docs/gett... Oh a
15. 315.▲
    by mkesper 3y ago

    I don't think so. You're right, that's awesome.
16. 316.▲
    by mkesper 3y ago

    No, just use an ad blocker and save the environment from more waste.
17. 317.▲
    by mkesper 3y ago

    Xppen has some really good graphic displays.
18. 318.▲
    by mkesper 3y ago

    Sadly, even the detailed specs don't tell how much unified memory (UMA) can be used. Would be quite relevant for running LLMs on the iGPU (like on Mx Macs). https://www.amd.com/en/product/14066 (AMD SmartAcce
19. 319.▲
    by mkesper 3y ago

    Scaling from zero did use quite some seconds so there would be some leeway for an alternative implementation.
20. 320.▲
    by mkesper 3y ago

    That's why optional GIL will be so important.
21. 321.▲
    by mkesper 3y ago

    But that's not a Macbook. And a Macbook M3Max with 128GB of RAM is almost 8000€.
22. 322.▲
    by mkesper 3y ago

    You can also specify format knowing about languages: https://git-scm.com/docs/gitattributes#_defining_a_custom_hu... Or even whole diff converters.
23. 323.▲
    by mkesper 3y ago

    This is a CP/M 2.2 installation, with the CCP loading from address D800h giving approximately 54Kb free for user programs. The system supports two 248K drives - Drive A is in ROM (and therefore readonly at present) and Drive B is in RA
24. 324.▲
    by mkesper 3y ago

    Modula2 was available and got used on Amiga. Silly teenager me found such high level languages "cheating" at the time.
25. 325.▲
    by mkesper 3y ago

    Distroless is Debian packages with worse security and without package management.
26. 326.▲
    by mkesper 3y ago

    That's like returning to publishing tarballs instead of having single, revertable commits.
27. 327.▲
    by mkesper 3y ago

    The generated code runs an order of magnitude faster than interpretation and 14% faster than LLVM -O0. An order of magnitude faster than interpretation. That's the interesting part for Python, I'd think.
28. 328.▲
    by mkesper 3y ago

    See Hainbach playing the room: https://m.youtube.com/watch?v=xm83hg7JXJ8
29. 329.▲
    by mkesper 3y ago

    You can buy multiple 4090s for that money and will get real GPUs including tensor cores. Still relevant it seems: https://timdettmers.com/2023/01/30/which-gpu-for-deep-learni...
30. 330.▲
    by mkesper 3y ago

    Thanks for posting the lwn.net link, reminded me of renewing my subscription!