#tyndur · 2009-12-10

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

  1. pcmattmantaljeth: did you end up getting a chance to review my patch?
  2. taljethI've had a look, yes, and we started a short discussion but quok had to leave
  3. pcmattmanAh. I unfortunately lost my backlog from last night, so I wasn't sure what was discussed
  4. taljethI'm not exactly happy with having CDI_DRIVER_CONST additionally because it's pure code duplication
  5. taljethSo if anything I'd give up the current CDI_DRIVER completely
  6. taljethXanClic and quok didn't even understand what your problem is, so no opinion from them yet ;-)
  7. pcmattmanYou mean, instead of the struct method, just use the explicit definition?
  8. taljethRight, although I really like doing it with the struct
  9. taljethBut having two declarations doing the same in each driver is just silly
  10. pcmattmanWell, the struct is cleaner
  11. pcmattmanBut it's also a level of indirection, which the compiler errors on in Pedigree modules
  12. taljethMaybe you can really have a try with the ugly macro? It should work and it's not code that you need to read each day
  13. pcmattmanI'll see what I can do
  14. taljethIt's not like I haven't thought about Pedigree. I would have liked to have also the name and the dependencies in the struct only, but this really wouldn't work for you, so I took something in between
  15. pcmattmanHmm, seems to work better than my original attempt ;)
  16. taljethWhich one?
  17. pcmattmanUsing the preprocessor to generate two functions for Pedigree's module code
  18. pcmattmanIt's also fairly neat, considering it's a preprocessor macro
  19. taljethRight, it must work good. It's my suggestion. ;-)
  20. pcmattman:)
  21. taljethI think it gets a bit uglier if you wanted to compile ata some day because that one defines two drivers in one file
  22. taljethBut you probably don't need it anyway
  23. pcmattmanYes, I noticed
  24. pcmattmanWe have our own ATA code that hooks deeply into the depths of Pedigree ;)
  25. FreakyPenguinSounds dangerous ;-)
  26. taljethpcmattman: Yes, we also have our legacy floppy driver ;-)
  27. pcmattmanThat said, if there is any situation where there ends up being more than one CDI_DRIVER call, I'll need to work some real magic into our CDI_DRIVER definition ;)
  28. taljethPreprocessor can be real fun if you don't need to do it every day ;-)
  29. taljeth*Preprocessor magic
  30. taljethI think basically you would register a module only once and have a second list of all drivers that you need to initialise in your wrapper
  31. taljethA bit more cumbersome, but still doable
  32. pcmattmanOkay, patch v2 is sent
  33. taljethHave you added the extern "C" to all header files now? I think we should it do right at once
  34. pcmattmanI'll add the remainder now and send as a separate patch
  35. pcmattmanThen I have to figure out how to call init_device for each driver's devices ;)
  36. taljethMaybe split your patch and have one patch with all the extern "C"
  37. taljethAnd another one with cdi-osdep.h
  38. pcmattmanGood plan
  39. taljethThe typedefs should at least get a cdi_ prefix. Also, I think in other place we don't have typedefs at all, but use the struct keyword. I won't insist on that, but for consistency it might be worth consideration
  40. taljethIs it even metadata? I think I'd call it something like cdi_pci_device_osdep when cdi_pci_device is the type that is "extended" by the struct
  41. pcmattmanWell, it could be information about the PCI device
  42. pcmattmanosdep works though, I'll change it
  43. taljethOk, thanks
  44. taljethI'll leave the rest of the comments to FreakyPenguin. It's getting late. ;-)
  45. taljethTalk to you tomorrow
  46. taljethOr rather this evening for you
  47. taljethGood night
  48. pcmattmanOkay, cya
  49. FreakyPenguingn8 taljeth ;-)
  50. pcmattmanThe patches are now on the list :)
  51. kwolfMoin
  52. DerHartmutMoin moin
  53. pcmattmanHi kwolf
  54. DerHartmutGood morning pcmattman
  55. DerHartmut-Good morning +Good evening
  56. pcmattmanMore patches on the list to review :)
  57. kwolfLooks like some work for the weekend :-)
  58. kwolfIf anyone wants to help me, you could review some of the patches and either comment on them or send your Acked-by
  59. DerHartmutI would do it, but I do not have any clue about CDI ^^
  60. quokHmm, anyone here want a google wave invite?