#tyndur · 2010-02-11

635 Nachrichten · Seite 1 von 4 · Original-CSV herunterladen

  1. pcmattmanMorning taljeth
  2. pcmattmanMore patches to make Pedigree's implementation even more obsolete? :P
  3. taljethGood morning pcmattman
  4. taljethNot only Pedigree, tyndur's implementation is now obsolete, too ;-)
  5. taljethSo we're converging at last ;-)
  6. pcmattmanHurrah :)
  7. taljethHm, it's pretty late... But the commit messages in #quokforge were kind of motivating to send this out today
  8. quokcommit messages are always fun :)
  9. taljethThough the CDI commits end up in no IRC channel
  10. taljethUnlike most other repos on git.tyndur.org
  11. pcmattmanWe probably need a #cdi, really
  12. taljethYet another channel?
  13. pcmattmanPerhaps
  14. quokthen we can abandon #tyndur! ;)
  15. pcmattmanActually, quok, now that you've popped up: any idea for an alternative name for cdi_mem_require_flags, that doesn't have anything to do with quok?
  16. pcmattman(ie, not cdi_mem_quok_area ;) )
  17. XanClicWhy not, I've joined 11 channels (10 right now), another one won't hurt ;-)
  18. quokno more cdi_mem_quok_area ? :(
  19. XanClicI love quok_area :D
  20. XanClictwo in favour of it, two against it -- let's listen to Freaky's and eddyb's opinions ;-)
  21. pcmattmanHmm, perhaps we should register CDI as a mentoring organisation for Google Summer of Code - we can get people to do work for us, they get paid, we get paid, everyone is happy
  22. pcmattman;)
  23. XanClicSounds nice ;-)
  24. XanClicBut it also sounds as I had to do something ;-)
  25. XanClicHm, bad English
  26. XanClicBut it also sounds as if I had to do something
  27. XanClicMaybe better
  28. taljethYou _have_ something to do ;-)
  29. pcmattmanWell, mentors have to supervise and help where needed the people doing the work
  30. XanClicArgh :D
  31. XanClicpcmattman: So no work for us except telling other people what to do? Nice ;-)
  32. taljethLike everyone else here. But seems I'm the only one actually working on the code. :-(
  33. pcmattmanBut it's $5000 USD for the participants, and only $500 for the mentors
  34. pcmattmanAnd only $500 if that mentor mentored a participant
  35. XanClicMoney is money ;-)
  36. quokI'm trying to think of a name for cdi_mem_require_flags, but coming up empty
  37. XanClicI think taljeth will be our mentor
  38. XanClicAnd we'll be the participants
  39. pcmattmanHaha
  40. taljethSo you want to get the money and I should go on with writing the code? ;-)
  41. XanClicHey, c'mon
  42. XanClicYou get some money, too ;-)
  43. pcmattmanhttp://socghop.appspot.com/document/show/gsoc_program/google/gsoc2010/faqs
  44. taljethXanClic: But then, it's only dollars
  45. XanClicHm, sometimes I have the feeling that one dollar equals to one euro
  46. XanClicAt least when buying American books
  47. pcmattmanWell, for the record, this year it's too late for projects to sign up
  48. pcmattmanBut next year... ;)
  49. taljethFor this year, quok needs to start a contest as a replacement ;-)
  50. XanClic*g*
  51. quokquok summer of code?
  52. quok:P
  53. pcmattmanHahaha
  54. taljethSounds good ;-)
  55. pcmattmanYay, my new 1 TB disk drive is being posted as we speak
  56. XanClicI don't speak, do you? ;-)
  57. pcmattmanShould be here on Monday, ready to help me get out of the problem where my current terabyte has mere megabytes free :(
  58. taljethXanClic: Better don't propose stupid ideas, someone might pick them up ;-)
  59. XanClic*g*
  60. taljethBut 1 TB... You need to write a lot of code to fill it up ;-)
  61. XanClicHm, I wonder how one is able to collect a TB of data
  62. XanClicI know how SHyx0rmZ does
  63. taljethYeah, that's what I though about the 230 MB disk then...
  64. XanClicHe collects Mangas (is that the same in English?)
  65. quokhow about cdi_mem_get_buffer instead of cdi_mem_require_flags? or cdi_mem_set_buffer_flags? cdi_mem_buffer_flags?
  66. XanClicbuffer_mem_flags_cdi? *g*
  67. XanClic(SCNR)
  68. quokhaha
  69. taljethflags_buffer_quok_cdi_mem?
  70. XanClicNice :D
  71. quokhey I through out some ideas :P
  72. quokthrew out, even
  73. taljethAh, was already wondering
  74. taljethHm, so far I still like cdi_mem_require_flags best, to be honest
  75. XanClicI think we should abandon useful descriptions. We could use /dev/urandom instead. Or just calling it cdi_create_signiphant. And winning the C obfuscation contest.
  76. quokwe should just use cdi_quok_function1 ;)
  77. quokfollowed, of course, by cdi_quok_function2 and cdi_quok_function3
  78. XanClicExactly
  79. taljethAnd cdi_quok_functionpi
  80. quokcdi_quok_functione
  81. XanClicAnd this function is inexact
  82. quokquok_function_alpha through quok_function_omega ;)
  83. XanClicThe more exact you give pi, the more exact the function will be executed
  84. quokhahahahaha
  85. XanClicCDI for Java2k
  86. quokcdi.net!
  87. XanClicThere's a probability of 0.0042 %, that the card sends total garbage? *g*
  88. XanClicgn8
  89. taljethpcmattman: Would cdi_mem_clone work better?
  90. taljethHowever, usually one would expect that a clone is exactly the same. But then, everyone knows that our code is broken.
  91. quokcdi_clone_quok ;)
  92. taljethYay, two quoks! :-)
  93. quokHmm I dislike cdi_mem_clone because it does imply that it'd be exactly the same
  94. quokwhich is part of why I suggest cdi_mem_get_buffer or something along the lines there. the get doesn't imply that the buffer may already exist or may be allocated or cloned or anything.
  95. pcmattmancdi_mem_make_sure_this_buffer_complies_with_the_flags
  96. taljethWhich is semantically the same as cdi_mem_require_flags, right? ;-)
  97. pcmattmanIt's more intuitive :P
  98. taljeth*lol*
  99. quokso we're going to rename is to cdi_mem_this_function_name_is_really_long_and_unintuitive_but_does_something_with_a_memory_buffer(), right?
  100. quoker, rename it*
  101. pcmattmanquok, you genius
  102. pcmattmanThat's the solution we've been looking for! :)
  103. taljethquok: I don't really like get_buffer for being unspecific, but it's definitely not insane
  104. quokpcmattman: troll quok is troll :)
  105. pcmattmanTrolling is fun ;)
  106. quoktaljeth: well, I was really just giving ideas for renaming cdi_mem_require_flags
  107. taljethI'm still not convinced that cdi_mem_required_flags is a bad name for it, but if you think so we might need to take a rather unspecific name like the one you proposed
  108. pcmattmanHmm, if only I could tell gcc to ignore a specific header file in a location
  109. pcmattmanAnyone here have direct commit access to the CDI git tree?
  110. pcmattmanHello kwolf :)
  111. kwolfYay, almost on time! :-)
  112. kwolfGood morning
  113. pcmattmanA bunch of ACKS and a PATCH on the list for you to peruse this morning/evening
  114. pcmattmanUnfortunately not the driver loading PATCH at this stage ;)
  115. pcmattmanBack soon
  116. kwolfHeh, I see what you want to achieve with that ;-)
  117. kwolfNot opposed to it, but can you include a paragraph on #include in doc.h? At first sight it looks counterintuitive, so it should be documented.
  118. kwolfHm, actually... We could change everything to angled brackets. It would use your version if it exists and the default one if you don't override it. Makes sense?
  119. pcmattmanBack
  120. pcmattmanYes, angle brackets would be preferable
  121. pcmattmanThey allow an OS to override a header if needed without modifying any of the CDI headers
  122. kwolfCare to send a patch which does the whole conversion?
  123. pcmattmanYes, I'll do that later tonight
  124. kwolfAnd for the other problem... Do you need to use cdi.git master for Pedigree or is it enough to have some commit that works, regardless of its branch?
  125. pcmattmanI'm fairly certain it requires master unfortunately
  126. kwolfHm... I don't want you to commit things into master without giving people a chance to comment on patches. But I would be fine with un-acked commits to a matt-devel branch or something like that.
  127. pcmattmanI don't have push access anyway
  128. kwolfThat's not the problem ;-)
  129. kwolfConfig files can be changed
  130. kwolfI can't do it right now because I don't have access myself from here, but if it helped you and we could still keep master clean, I would do it this evening.
  131. pcmattman"Right
  132. pcmattmanI don't disagree with requiring patches to be acked, mind you
  133. pcmattmanThis should be one of the last cases where something in CDI actually causes the Pedigree build to fail, because I'm going to be watching new patches closely now :)
  134. kwolfWe're using this acking policy for quite a while now in tyndur, and surprisingly the quality of commits has improved a lot. ;-)
  135. pcmattmanMaybe I should adopt the same workflow for Pedigree ;)
  136. kwolfSo I'm not going to give it up, and I'm glad that you're with me :-)
  137. eddybi wonder how many NAKs i'd get(as if the USB ones weren't enough) :P
  138. kwolfhehe
  139. kwolfWell, that's part of the process
  140. pcmattmanNAKs are better than having a commit reverted ;)
  141. kwolfI've been sending [PATCH v5] mails myself. But better the bugs are caught before the commit.
  142. eddybokey, matt, then
  143. eddybit will take a while to check it :P
  144. eddybcould we use the forum for patches?
  145. eddybmailing lists are harder to read :P
  146. kwolfThen fix your mail client ;-)
  147. eddybmail client?
  148. eddybfirst, i don't have/use one
  149. eddybi mean, i use webmail
  150. pcmattmanUse mutt ;)
  151. kwolfeddyb: wtf?
  152. pcmattmanYou'll learn man at the same time
  153. eddybi was reffering to the mailing lists you find if you search for patches :P
  154. kwolfYou're referring to mailing list archives, I guess
  155. eddybyeah, that's how they are called
  156. kwolfActually, the main use case of mailing list is - surprise - to get mails ;-)
  157. eddybi ussually get lost
  158. eddybhmm
  159. eddybif it's not a long discussion, i may be able to follow it :P
  160. eddyboh, shet
  161. eddybit always helps to right-click :P
  162. eddybthat's where the spell-checker was hiding
  163. eddybnot in the options XD
  164. pcmattmanRaincheck on the patch in case I don't get it done tonight ;)
  165. kwolfAs usual ;-)
  166. kwolfpcmattman: But can you check if a branch would work? I mean, in the end it's only a hash of a commit that you use for the submodule checkout, so in theory it should be possible somehow.
  167. pcmattmanHopefully
  168. pcmattmanTheoretically it should
  169. pcmattmanHopefully the git developers didn't think "let's hope nobody needs this"
  170. kwolfHm, do we have any branches in cdi.git? I think there must be some old one
  171. pcmattmanI wouldn't have gotten remote branches in the original clone
  172. pcmattmanIt would've just grabbed master
  173. kwolfOh?
  174. pcmattmanI'm not sure how git submodules differ from conventional git repositories
  175. kwolfBut you can try to get the usb branch checked out as a submodule. If it works, we'll do it this way.
  176. pcmattmanI'll have to have a read of the man pages
  177. kwolfI don't think there is any real difference for the submodule repository
  178. pcmattmanAt least git's man pages aren't too hard to read
  179. kwolfBut then, submodules are still magic for me ;-)
  180. pcmattmanI still don't quite understand how they work ;)
  181. pcmattmanKeep the magic in the box and pretend it's not there seems to work
  182. kwolfWe might be coming to a point where it doesn't work any more
  183. pcmattmanWhy's that?
  184. pcmattmanI find it a handy way to (pretend to) keep up-to-date with the latest version of all the drivers and CDI headers
  185. kwolfI mean preteding it's not there might not work any more when we need to work with remote branches
  186. kwolfAnyway, I just need to know if a branch works for you or doesn't. You'll need to try it with the existing usb branch.
  187. pcmattmanAh
  188. pcmattmanYeah, we definitely have to experiment with it
  189. pcmattmanWorst case is that it doesn't work at all and I just have an additional CDI working copy that I copy in and out of all the time
  190. pcmattmanI'd symlink if I weren't on Windows...
  191. kwolfpcmattman: That's what I do. It's not very convenient.
  192. kwolfBut maybe I should have a look at subtree merges once again
  193. kwolfLooks like even more magic than submodules, though ;-)
  194. pcmattmanSubtree merges sound like black magic
  195. pcmattmanMight want to write your root password and put your ssh keys into your will if you go down that route ;)
  196. kwolf*lol*
  197. kwolfBeing able to cherry-pick forth and back would be exactly what I want
  198. kwolfBut I'm not sure if it's doable
  199. eddybmoin, XanClic
  200. XanClicMoin, moin