Chat Logs

  1. [twisti]databases ruin you for the world
  2. [twisti]there isnt a UI i can walk by without wishing for free where/order by input
  3. LaSombraI'd kill for any website that shows a list of things to have a DSL for querying
  4. dreamreali wonder what that would look like
  5. Para0:0:0:0 TCP/5434
  6. Para5432 even
  7. dreamrealpffft, I'm serious
  8. ParaAll tabular views should init themselves from list of objects -> autoinject per-column filters and sorts.
  9. dreamrealThat's one optio. But would humans use it?
  10. dreamrealLike, actual humans?
  11. dreamrealMaybe it doesn't even matter, hmm, maybe I'm thinking of the problem the wrong way
  12. ParaSolution looking for a problem.
  13. ParaThat would work if one could somehow indicate such lists on page.
  14. ParaAnd then pattern apply the filter selector thingy on it.
  15. dreamrealWell, the website's raison d'etre has to come into play somewhere
  16. deebojust solve paging with offset and fetch n rows only
  17. deebodoes any rdbms support some sort of session / cookie for this? or some sort of cursor that can be navigated both ways without holding on to a connection for the duration
  18. ParaIt's a query strategy.
  19. ParaYou need to support temporal queries for stability, apply sorts, do offsets.
  20. dreamrealdeebo: you kinda need state as part of the process right now
  21. dreamrealso you can treat a website as a db-ish thing, semweb tried it, sort of
  22. dreamrealhave any of you used semweb?
  23. dreamreal(the question is its own answer)
  24. deeboi meant mostly to avoid overhead from offset scrolling to nth page
  25. dreamrealoffset scrolling isn't expensive in a decent db
  26. dreamrealdunno about sqlite, but pretty much every other db is fine
  27. ParaSELECT * FROM article WHERE valid_from <= $timestamp AND valid_until > $timestamp ORDER BY valid_from DESC, <other clauses> LIMIT 20 OFFSET 40; -- or whatever is your pagination size and offset etc.
  28. ParaAnother strategy is to cut queried results based on last id from previous page set and then taking next n results based on page size.
  29. dreamrealThe good news about that approach is that you've limited your users to the four who understand SQL *and* your site's data model.
  30. ParaGood, no wasted resources on idiots,.
  31. dreamrealand the site better be a passion project for the author forever, because it's not going to get monetized unless it's selling something, which makes the model untenable in the first place :D
  32. deeboif you can get index coverage for any combination of filters to avoid table scans it's fine
  33. deebogimme lucene/(open|elastic)search "tables" in my postgres
  34. dreamrealgood news, you already have them :D
  35. dreamrealif you're using postgres you really don't need lucene except for the query language, if that's useful to you
  36. deeboi tested some text search thing with one "backoffice" solution, but a covering index ended up being better and easier to reason about
  37. dreamrealsure
  38. dreamreallucene, solr, ES are all good bridges to scanning for data; if you really wanted to query it deeply you'd use sparql and rdf, but there's a reason people know about sparql and rdf and they don't know sparql and rdf themselves
  39. computerscienceshould I use java or node
  40. julemand101computerscience: For what problem?
  41. computersciencegetting a graduate developer job in sydney
  42. julemand101I mean, would it not be better you do a research into opening job offerings in your area to see what they look for?
  43. ParaOr if you're interested in some specific field, one might be generally favored over another.
  44. ParaOr if your interests match better with specific ecosystem.
  45. dreamrealcomputerscience: depends on for what, but honestly: node
  46. computersciencewhich node framework
  47. dreamrealnextjs
  48. computerscienceI thought express was better
  49. computersciencemy university taught us express
  50. computersciencemaybe because it's basic
  51. dreamrealdude, you're in #java, asking a question like that. I don't care what you use. If node is an option, the mechanics are broken. Use something else, not our stuff.
  52. dreamrealnode and java are not on the same playing field; it's like saying "I have the choice between an f1 and a ford truck. Which one shoudl I use?"
  53. dreamrealThe question itself is a category error.
  54. dreamrealYou wouldn't want an f1 car if you're carrying a load of cordwood. You wouldn't use an F150 on an f1 track.
  55. dreamrealif the f150 is an option, then gosh, make the choice and live with it. Node is easy. Asking the question makes it sound like that's a viable metric to use for you. (Java's easy, too, but in completely different ways.) Use what works for you.
  56. NeXeNwhy does bytebuddy throw errors and still build properly ?
  57. NeXeNi mean, i see stack traces, but it's just a debug thing in hibernate, but wtf, like i always wondered about that
  58. NeXeNit offends my sense of ocd
  59. jreicherThat is curious. What's the error?