ByteCode.News
RSS
SubmitLog in

Community

Chat Logs

Sunday, March 22, 2026

  1. phalethtime to say farewell to the edge runtime https://images4.imagebam.com/95/eb/b9/ME1BIGIR_o.png it's memory consumption started at 137M over a day and half ago
  2. bluephaleth: lol
  3. bluephaleth: https://github.com/primate-run/primate/issues/262
  4. nevetProposal: Introduce `p.uuid()` as a first-class Pema type 路 Issue #262 路 primate-run/primate
  5. phalethI think mysql supports uuids and turns them into 16 byte long binaries
  6. phalethwhich it can then easily compare and so use as primary keys
  7. phalethnot sure why it's a 12-byte length for MongoDB, sounds odd
  8. phalethah, it's not UUID, it's some mongo objectid thing
  9. phalethin any case, having database compare strings/texts is a horrible idea
  10. phalethand also ordered is always much much better, that's already there in the proposal
  11. blueyes, phaleth
  12. bluethis mostly closes a gap I've been annoyed with for a while
  13. blueand mongo ObjectId is some proprietary nonsense, but it's the most effective way to do primary indexing in mongo
  14. blueI'm incling to say that p.uuid (datatype: "uuid") will map to that, while the bespoke types "uuid_v4" and "uuid_v7" will use uuids,which mongo db cannot effectively index, so perhaps we'll add a warning for that in the mongodb driver
  15. blue"uuid" as a datatype should mean, "use the best UUID approximation you have". sqlite doesn't have one, it's just TEXT. so it's logical that for mongodb, that would be ObjectId.
  16. bluephaleth: do you mean this?
  17. bluehttps://dev.mysql.com/doc/refman/8.4/en/miscellaneous-functions.html#function_uuid-to-bin
  18. phalethyeah, if uuid is used as primary key, mysql and all the other dbs that support uuids should automatically sort records by that primary key after insertion
  19. phalethI guess then for mongodb you will have to make an exception
  20. phalethbut overall the idea sounds good
  21. bluephaleth: https://github.com/primate-run/primate/issues/262#issuecomment-4106146943
  22. nevetProposal: Introduce `p.uuid()` as a first-class Pema type 路 Issue #262 路 primate-run/primate
  23. bluethat's a follow-up with your mysql idea
  24. blueclanker also ack'd, so looks good
  25. bluephaleth: can I restart the vps? arch has linux 6.19 out
  26. phalethyeah, sure
  27. phalethbye bye blue
  28. bluephaleth: well, it's restarted, but primate.run is dead :(
  29. blueoh, now it's working
  30. blueI guess it took the haproxy podman container some time to boot up
  31. phalethsystemd is prolly slow
  32. blueya
  33. phalethhaproxy is on the host OS
  34. blueit's a podman container though, right?
  35. phalethhaproxy is installed on archlinux host OS directly
  36. phalethnot a container
  37. blueoh, gotcha
  38. blueI thought it used to be an nspawn container
  39. phalethit used to be, yeah
  40. blueis it easier to have it on host?
  41. phalethit's just a bit more convenient
  42. phaleththere is less NAT translations happening when haproxy is not in container
  43. phaleththe reason I've put haproxy into container was that I used to do custom builds
  44. phalethbut the archlinux build of haproxy has everything needed
  45. phalethso there is no reason to slow things down anymore
  46. phalethand also of course archlinux or whoever maintains the haproxy package makes sure it's up to date
  47. bluegotcha
  48. bluemakes sense
  49. phalethhaproxy's setup is mostly about it's configuration (haproxy.cfg) and that one is exactly the same as it used to be
  50. blueya
  51. bluein the future I do intend for servers to regularly update and restart, but that's a load-balancing issue
  52. blueyou shortly reroute all machines on server 1 to server 2 (assuming you got redundancy on server 2), update server 1, put them back, then update server 2
  53. phalethheh, ok