Public Pastes
BitBucket - Backup your code in the cloud! Host unlimited private projects, for free. SIGN UP takes 10 seconds, and it's free!
SHARE
TWEET

more Database worries

a guest Jan 31st, 2016 6 Never
  1. 10:22:58    ~copypaste  i'm trying to find the schema again, it's clever
  2. 10:23:06    ~copypaste  again, "clever" :^) so you know where i am going with this
  3. 10:23:13    n-tech  yeah fuck, god damnit. This is raw SQL.
  4. 10:23:26    n-tech  I'm really, really concerned about 'clever' solutions. Right now it's a clean slate. Everything is in order.
  5. 10:23:36    n-tech  I'm worried about a situation where we're just like, oh lets borrow all these old assets
  6. 10:23:42    n-tech  and end up with the same thing with a different name
  7. 10:24:01    ~copypaste  doesn't matter if it's clean if it doesn't work. also we have never used KX schema before btw. i didn't know it already has this problem solved until informed of such fact by n-tech
  8. 10:24:05    ~copypaste  no, not n-tech. czaks
  9. 10:24:09    ~copypaste  sorry, mixing shit up in m head.
  10. 10:24:19    ~copypaste  i never used KX on any live site.
  11. 10:24:34    n-tech  it doesn't matter if it works if it's built like shit
  12. 10:24:40    ~copypaste  i tried it once, but only to help STI with a migration
  13. 10:24:53    n-tech  it has to be done right. it's worth taking time to investigating every issue to see how the convention is done.
  14. 10:24:59    ~copypaste  ok. but we need to rethink the schema. to be honest n-tech, let's be real. 2ch's system is best.
  15. 10:25:01    ~copypaste  OP is always >>1
  16. 10:25:04    ~copypaste  first reply is always >>2.
  17. 10:25:07    ~copypaste  and so on and so forth.
  18. 10:25:20    ~copypaste  that is the most sense to the DB.
  19. 10:25:23    n-tech  for every thread? are you sure you'd prefer that?
  20. 10:25:29    ~copypaste  no, they won't. but that makes the most sense.
  21. 10:25:43    ~copypaste  you need to understand that the IB use case goes against what the databas wants
  22. 10:25:48    n-tech  it does, the board_id is literally an unintended side effect of having a table per board.
  23. 10:25:48    ~copypaste  that's why STI did it the separate tables
  24. 10:26:06    ~copypaste  n-tech: there ya go.
  25. 10:26:12    ~copypaste  and we've been carrying that since 2003.
  26. 10:26:19    ~copypaste  it's far too late to change it and is too much of a shock
  27. 10:26:36    n-tech  oh I thought you were talking about moving to that system
  28. 10:26:36    ~copypaste  look at this thread
  29. 10:27:04    ~copypaste  i am saying that it is important to have this feature, even if it requires some raw SQL triggers or otherwise.
  30. 10:27:10    ~copypaste  we can write one for mysql and one for postgres.
  31. 10:27:25    ~copypaste  most large web applications have some raw SQL. i am sure that eloquent even has a Eloquent(  ??? )ql() or similar.
  32. 10:27:32    ~copypaste  oh, nice, my emote kicked in
  33. 10:27:33    ~copypaste  lol
  34. [...]
  35. 10:27:52    n-tech  I really don't want to halfass anything right now. anything done poorly will haunt us for 10 years
  36. 10:28:05    n-tech  we'll think back on this shit like "fuck why did we do x we're retarded"
  37. 10:28:05    n-tech  lol
  38. [...]
  39. 10:29:42    ~copypaste  so this is the DB basics. let me install kusaba and pull their schema
  40. 10:29:53    n-tech  but my point is that right now, we have a setup where there cannot be a duplicate board id. that means it'd be rejected
  41. 10:29:55    ~copypaste  and i can tell you how they do it. i know for a fact it is not triggers and it exploits an undocumented "feaeture"
  42. 10:30:06    ~copypaste  n-tech: on fast boards, like /pol/ during a chimp out
  43. 10:30:09    ~copypaste  that's reject half of people's posts
  44. 10:30:12    n-tech  but
  45. 10:30:15    n-tech  we can loop it
  46. 10:30:17    ~copypaste  you think i'm kdding, but our current post.php already has race conditions
  47. 10:30:19    n-tech  if it's rejected for a duplicate ai
  48. 10:30:21    ~copypaste  how is that not half assed?
  49. 10:30:22    n-tech  we can try again
  50. 10:30:22    ~copypaste  lmao
  51. 10:30:40    ~copypaste  while ($noerror) tryagain(); ?
  52. 10:30:41    ~copypaste  :^)
  53. 10:30:44    n-tech  that seems more correct than breaking the database agnosticism almost immediately
  54. 10:30:45    ~copypaste  what could possibly go wrong?
  55. 10:31:17    n-tech  make it more clever. if the number of posts didn't change, bail out with an error. if it did actually go up you know a concurrency issue occured.
  56. 10:31:46    ~copypaste  i understand your concern with the database agnosticism, i really do
  57. 10:31:56    ~copypaste  but i also know incrementing IDs outside of the RDBMS is pure cancer
  58. 10:32:15    ~copypaste  i have worked on threre projects that did it, and it caused problems every single time.
  59. 10:32:23    ~copypaste  i did fix one of them with a loop in fact.
  60. 10:32:35    ~copypaste  that's not a fix, it's a hack. if the SQL DB is waiting on posts
  61. 10:32:40    ~copypaste  one attempt can try 10 times.
  62. 10:32:47    ~copypaste  you think this isi a joke but i can easily write a test case.
  63. 10:33:05    ~copypaste  database agnostic ORMs are supposed to be helpers for common things like select, insert, delete, join
  64. 10:33:06    n-tech  I'm not saying it's a joke, I'm just extremely worried about reducing the available featureset of the framework we've adopted immediately
  65. 10:33:09    n-tech  to whatever can support this hack
  66. 10:33:09    ~copypaste  not excuwses to never know SQL
  67. 10:33:28    ~copypaste  i think if we find a way to support it in sqlite, postgres, mysql
  68. [...]
  69. 10:33:57    czaks   this case for mysql we talked about?
  70. 10:34:00    ~copypaste  yes
  71. 10:34:06    czaks   myisam and aria support that
  72. 10:34:06    ~copypaste  remember this discussion? i still have the log
  73. 10:34:10    ~copypaste  oh yeah?
  74. 10:34:13    ~copypaste  just innodb doesn't?
  75. 10:34:13    czaks   yes
  76. 10:34:17    czaks   and innodb doesn't
  77. 10:34:20    ~copypaste  great. so we can just reject innodb.
  78. 10:34:33    ~copypaste  Does Eloquent even let you pick?
  79. 10:34:37    ~copypaste  i guess when you CREATE DATABASE
  80. 10:34:43    czaks   just that aria is only in mariadb
  81. 10:34:48    ~copypaste  that's fine.
  82. 10:34:53    czaks   and you rejected mariadb back in the times xD
  83. 10:34:56    ~copypaste  fuck mysql. i install mariadb and we save 50% CPU
  84. 10:34:59    ~copypaste  haha :DDD
  85. 10:35:04    n-tech  I'm sure there's a way to not support specific schemas, but really? we're not going to support innodb?
  86. 10:35:05    ~copypaste  it wasn't so popular back then
  87. 10:35:10    czaks   because muh stable software
  88. 10:35:22    ~copypaste  as i said, it wasn't so popular back then, mariadb
  89. 10:35:27    czaks   i see
  90. [...]
  91. 10:37:46    ~copypaste  i think the main conflict is that i wouldn't mind doing that sometimes, where n-tech is opposed to ever writing SQL
  92. 10:37:47    ~copypaste  :DD
  93. 10:38:08    ~copypaste  it's this
  94. 10:38:09    ~copypaste  http://laravel.com/docs/4.2/eloquent
  95. 10:38:32    n-tech  5.0
  96. 10:38:38    n-tech  4.2 is the old version, be careful of reading old doc
  97. 10:38:41    ~copypaste  oh
  98. 10:38:45    ~copypaste  you already adopted 5
  99. [...]
  100. 10:49:26    n-tech  it actually makes me sort of nauseous to think
  101. 10:49:35    n-tech  that we're going to have to code in this huge fucking ordeal
  102. 10:49:38    n-tech  to support some autistic bygone
  103. 10:49:42    n-tech  fuck man
  104. 10:49:44    ~copypaste  i know. i'm sorry.
  105. 10:49:54    ~copypaste  but changing the numbers would destroy autism
  106. 10:50:21    @linear     i dont even want to know what is happening
  107. 10:50:31    ~copypaste  linear: we're rewriting 8chan
  108. 10:50:33    ~copypaste  infinitydev.org
  109. 10:50:35    ~copypaste  (? ????)??
  110. 10:50:42    @linear     how much of it
  111. 10:50:44    [Desu]  trying to prevent patrick bateman from crying by the sounds of it
  112. 10:50:45    n-tech  lol
  113. 10:50:46    ~copypaste  everything.
  114. 10:50:48    n-tech  just the tip linear
  115. [...]
  116. 10:51:06    ~copypaste  we are quite literally trying to fix dubs right now
  117. 10:51:07    [Desu]  hehehe
  118. 10:51:12    ~copypaste  i'm not even joking
  119. 10:51:13    [Desu]  hheheheheh
  120. 10:51:15    [Desu]  glorious
  121. 10:51:16    ~copypaste  this whole argument is over dubs/trips/quads.
  122. 10:51:17    ~copypaste  lo
  123. 10:51:18    ~copypaste  l
  124. 10:51:20    n-tech  just fuckig tink
  125. 10:51:22    n-tech  of a world
  126. 10:51:26    [Desu]  Patrick Bateman smiles upon you this day
  127. 10:51:30    n-tech  where udina doesn't shitpost every 100,000 post
  128. 10:51:35    n-tech  do you want to live in that world
  129. 10:51:38    [Desu]  gno
  130. 10:51:59    n-tech  i guess /udina/ is more of a pissposter tho
  131. 10:52:09    [Desu]  /undina/?
  132. 10:52:19    n-tech  that's the one
  133. [...]
  134. 10:55:48    n-tech  ; ; i still think my loop hack is better than requiring certain databases
  135. 10:55:49    ~copypaste  right
  136. 10:55:53    n-tech  i am a simple man with simple concepts
  137. 10:56:01    [Desu]  it gives a good idea if the board is active too.
  138. 10:56:07    ~copypaste  the loop hack is awful and puts unneeded strain on the db
  139. 10:56:15    ~copypaste  please talk sense in to him linear i need to go away for a few mins
  140. [...]
  141. 11:02:26    ~copypaste  n-tech: i know this is autistic but they won't want to use it if the numbers aren't right
  142. 11:02:35    ~copypaste  i am 100% certain of that
  143. 11:02:50    ~copypaste  so i am in favor of drop innodb.
  144. 11:02:54    ~copypaste  and use KX like schema.
RAW Paste Data
Top