#tyndur · 2009-08-15

170 Nachrichten · Seite 1 von 1 · Original-CSV herunterladen

  1. taljethMoin
  2. FreakyPenguinMoin
  3. taljethMoin FreakyPenguin
  4. taljethFreakyPenguin: Are you ok with my changes to the logs page?
  5. FreakyPenguinDoesn't look too bad ;-)
  6. taljethKonqueror doesn't render the border for the <select>. KRITIKEL! ;-)
  7. FreakyPenguinWell... fix it! ;-)
  8. taljethMaybe when you have finished porting KDE to tyndur :-P
  9. FreakyPenguinBut that's okay for me, as long as w3m renders it properly. ;-)
  10. taljeth*hehe*
  11. taljethHm, is there anything useful to do? I definitely don't feel like fixing the git port...
  12. pcmattmanHmm, trying to run my module... I get "Befehl wurde nicht gefunden"
  13. taljethWhich is "command not found" as you probably have guessed
  14. taljethDoes the file exist?
  15. pcmattmanHmm, it does
  16. pcmattmanBut I'm using a US keyboard layout, so I need to use SHIFT-7 to get a proper slash... Oops :)
  17. taljethHeh, you'll not only learn the German language but also the German keyboard layout, it seems ;-)
  18. SHyx0rmZhello
  19. taljethHi SHyx0rmZ
  20. FreakyPenguintaljeth: What about documenting cdi? I dont feel like debugging ata on lio2 neither... ;-)
  21. taljethpcmattman: Does it work?
  22. taljethFreakyPenguin: But CDI is perfectly documented already? ;-)
  23. taljethIs the current LIOv2 code in the git?
  24. pcmattmantaljeth: It does. Now I just need to write the driver... :)
  25. FreakyPenguin*g*
  26. taljethpcmattman: Sounds like the easy part ;-)
  27. pcmattmanHalfway through device initialisation now, so I'm making good time :)
  28. FreakyPenguintaljeth: I don't think I commited the kernel ata already...
  29. FreakyPenguinBut I think I could change that. ;-)
  30. taljethRight, last commit is from the last day of hackathon
  31. pcmattmanBack soon, dinner :)
  32. taljethGuten Appetit (don't think that exists in English?)
  33. FreakyPenguintaljeth: Now I've pushed some changes. I hope it compiles... ;-)
  34. FreakyPenguinWith this code I'm able to mount ext2, but the root dir stays empty. iirc ext2 doesnt get correct data when it tries to read the root inode.
  35. taljethWhich means that you'd better use some simpler tests
  36. taljethFreakyPenguin: Do you have any plans what to do with the comments in CDI driver code, regarding their language?
  37. taljethI guess we can have the public headers multilingual, just like bluecode does in his C++ library. But for inline comments this feels ridiculous.
  38. FreakyPenguinSo there's only one option I think...
  39. taljethpcmattman: Or how do you feel about it?
  40. taljethI suspect a standard containing "The driver author SHALL learn German" isn't a solution ;-)
  41. FreakyPenguinBilingual headers would be nice, if we can keep the comments in sync... ;-)
  42. pcmattmanI think driver code can be in whatever language, as the idea is that drivers shouldn't need edits to be ported
  43. taljethI think we can. pcmattman just need to add his NACK to each patch modifying the German version, but not the English one ;-)
  44. taljethpcmattman: In general, yes. But let's take your floppy driver patch as an example. Should I translate your additional comments? Having mixed-language comments seems the worst of all options to me.
  45. pcmattmanI would say each driver would have a specific language and stick to it
  46. pcmattmanSo you would probably end up translating my comments
  47. pcmattmanThe headers though I think should be bilingual, to make it easy for someone who doesn't know German to learn what a function does
  48. pcmattmanThat's just my 2 cents though
  49. taljethSure, I couldn't agree more for the headers
  50. taljethIt should be easy enough to even strip the "wrong" language when copying the header for a specific OS
  51. pcmattmanPerhaps
  52. taljethWould make merging later on more difficult, though
  53. pcmattmanIdeally though you'd want at least the headers to be directly portable
  54. pcmattmanThat way porting CDI to a new OS is just implementing functions rather than modifying header files
  55. taljethYou shouldn't modify the header files ever (except when adding new functions to CDI itself). That I'm doing it all the time is no excuse. ;-)
  56. FreakyPenguin*g*
  57. bluecodehi there
  58. FreakyPenguinHi bluecode
  59. taljethMoin bluecode
  60. taljethpcmattman: Btw, I think we still need to find a way to use CDI drivers in Pedigree without patching them. Or have you already managed to do this?
  61. bluecodewhy do you (have to) patch them for pedigree?
  62. taljethThe first reason is that I implemented it wrong ;-)
  63. bluecodewhat specifically?
  64. taljethThe second is, you need things like MODULE_NAME() and so on in each driver
  65. bluecodewhich seems like a good idea anyway,doesn't it?
  66. taljethbluecode: I added a Pedigree specific field to struct cdi_device. Which was okay for a quick hack, but definitely can't stay this way
  67. taljethbluecode: Agreed. It's just not there yet
  68. pcmattmantaljeth: I ended up making a "pedigree.cc" file which you add to the driver directory and modify to set the right name
  69. FreakyPenguinI just added a cdi-devel list to list.tyndur.org. Hope it works ;-)
  70. taljethDo I need to subscribe myself or have already subscribed a standard set of interested people? ;-)
  71. FreakyPenguinNo, till now I don't have chosen any volunteers ;-)
  72. FreakyPenguinOkay, you are subscribed
  73. bluecodepcmattman: what C++ classes are allowed to be used in pedigree modules? Nothing from the standard library, right?
  74. pcmattmanbluecode: The full kernel functionality is exposed to modules, but no standard libraries
  75. taljethWhich reminds me that we need to get rid of the printfs in drivers
  76. bluecodepcmattman: so there is probably not much ground for a C++ interface to/for CDI...
  77. pcmattmanNot unless we ported the STL to Pedigree kernel space...
  78. taljethbluecode: You need volunteers to help you with the lightOS CDI interface? ;-)
  79. bluecodeyou mean implementing CDI for lightOS?
  80. bluecodeor a C++ interface?
  81. taljethThe C++ interface that you're probably going to use for implementing CDI for lightOS?
  82. bluecodeI don't think I need C++ to implement the CDI stuff
  83. taljethThen I guess my assumption was wrong
  84. bluecodewhat's backdev in struct cdi_device for?
  85. taljethSomething with ATAPI
  86. taljethI think the block driver uses this to reference the SCSI driver or something like that
  87. bluecodewhy is it tyndur-specific then?
  88. bluecodeat least it says so in the comment
  89. taljethI guess because nobody cared to standardize it ;-)
  90. taljethjanosch had a different implementation of ATAPI, more or less outside CDI
  91. taljethFreakyPenguin: You should have some ne2k hardware around, right?
  92. bluecodeshould it be permittet to write a driver in C++ (only calling the current CDI functionality)?
  93. FreakyPenguintaljeth: ISA?
  94. taljethPCI
  95. FreakyPenguinI dont think so...
  96. bluecodebochs has one
  97. pcmattmanPCI
  98. bluecodebut tyndur on bochs is fun *g*
  99. taljethqemu has one, too
  100. taljethBut real hardware usually behaves differently
  101. bluecodesure
  102. taljethAs your pcnet driver proves ;-)
  103. FreakyPenguinI have a bunch of ne2k ISA, but I dont think I have any pci-versions of them ;-)
  104. taljethHm, ok. stultus might have one
  105. taljethAt least he has a 3c90x for the next driver ;-)
  106. bluecodeI asked the question cause extern "C" might be missing
  107. pcmattman#ifdef __cplusplus extern"C" {..... is missing from all the headers
  108. bluecodeexactly
  109. pcmattmanThat's not a difficult change though
  110. bluecodeof couse not, I am just asking if it is supposed to be there
  111. taljethI'm not exactly thrilled about C++ drivers. You wouldn't want Pascal drivers either. But if it helps in any way I wouldn't oppose the change.
  112. taljethI guess it helps when writing the OS specific CDI implementation, too, so it would make sense
  113. taljeth(und sogar welchen ergeben *g*)
  114. FreakyPenguin*g*
  115. bluecodeI am not going to giggle over this, language nazi ;-)
  116. taljethThose Germans don't have any humour *sigh* ;-)
  117. bluecodeof course not!
  118. bluecodeshould I send a patch with the extern stuff?
  119. taljethI don't need it, but you and pcmattman might profit. Feel free.
  120. cwo_pengen_Gtuanheh nice software here http://www.sharehaha.com/download.php?id=5C804A041
  121. taljethpcmattman: You definitely need an editor that highlights trailing whitespace ;-)
  122. pcmattman:(
  123. pcmattmanOne day I'll create a patch that is correct the first time. That'll be a good day.
  124. FreakyPenguin*g*
  125. taljethHeh, we'll see. ;-)
  126. taljethNo need to resend, though, I can fix it myself
  127. taljeth|ne2kSeems to work so far
  128. FreakyPenguinNice :)
  129. taljethAnyone feeling like reviewing the code? ;-)
  130. pcmattmanHmm, internet dropped out...
  131. taljethHope you're not using our network drivers ;-)
  132. pcmattmanHehe, just Windows drivers :P
  133. taljethHm, is the tyndur coding style applicable to CDI drivers?
  134. FreakyPenguinI think this should be defined for each driver...
  135. pcmattmanI tried to uphold the coding style used by the RTL8139 driver
  136. taljethIt's only minor details like the space after if or uint16_t* foo vs. uint16_t *foo
  137. taljethI think I'll just commit it as is
  138. taljethWe already have janosch's code and it deviates from our coding style much more
  139. taljethpcmattman: I'm wondering... You haven't complained one single time about aynthing in CDI now. Is it so easy or are you just being polite? ;-)
  140. pcmattmanI honestly have not had a problem with it so far
  141. FreakyPenguinSo you didn't use cdi/fs, right? *g*
  142. pcmattmanHehe, correct
  143. taljethHeh, right. I prefer not to touch cdi/fs if it can be avoided.
  144. pcmattmanWe already have FAT, ext2, and ISO9660 drivers. So I didn't really think much about cdi/fs
  145. taljethThere's probably no point in it then
  146. taljethAt least as long as nobody starts to write a btrfs driver or so
  147. taljethOr maybe NTFS
  148. pcmattmanIf there ends up a CDI NTFS driver, I'll probably end up putting cdi/fs stuff into Pedigree :)
  149. taljethWell, I don't think I'm going to write one. I don't have any NTFS partitions.
  150. taljethAre you going to go beyond network drivers and floppy for CDI in Pedigree?
  151. taljethI'm thinking of planned things like USB, sound or video (well, for the latter James have already stated that you are going to do your own thing)
  152. pcmattmanUSB we already have code that's going to be reviewed and integrated soon
  153. pcmattmanBut definitely sound, if it has a good interface
  154. taljethWe need to define it yet
  155. pcmattmanYup
  156. taljethWith USB, looks like bad timing :-(
  157. pcmattmanAdditionally we want to tackle USB with C++, because our design of the driver(s) shows that there's a clean inheritance pattern
  158. FreakyPenguinWell, you can do inheritance with C too. *g*
  159. pcmattmanI don't believe in trying to use object-oriented practices in non-object-oriented languages :)
  160. pcmattmanLots of people disagree with me, but that's just how I see it
  161. taljethBasically you're disagreeing with CDI then
  162. pcmattmanPerhaps
  163. pcmattmanI might rephrase...
  164. pcmattmanI don't believe in trying to use *excessive* object-oriented practices in non-oo languages
  165. taljeth*g*
  166. pcmattmanI don't like seeing projects where they *clearly* should be using C++
  167. taljethIt's always the question where you draw the line
  168. pcmattmanAnyway, I'm going to sleep now. Night all
  169. FreakyPenguingn8 pcmattman
  170. taljethgn8