Announcement

Collapse
No announcement yet.

ATI R600g Gains Mip-Map, Face Culling Support

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • On the other hand, having a test which uses the API in somewhat incorrect way is not bad either.

    Comment


    • Originally posted by marek View Post
      On the other hand, having a test which uses the API in somewhat incorrect way is not bad either.
      Assuming the correct behaviour is to fail the test, sure.

      Comment


      • Originally posted by nanonyme View Post
        Assuming the correct behaviour is to fail the test, sure.
        Hm, let's say "pass the test by failing"

        Comment


        • I might open a new thread, but I came to think of the following:
          In my vdpau backend, I am using some bits from the mplayer implementation to my bitstream parser. Well i know mplayer is GPL'ed and my code will be too, but do I have to copy any kind of license from the GPL'ed source file from mplayer into my code?

          Comment


          • I'm not a lawyer but I'd suspect loaning ideas from there is a just fine without any license concerns but loaning code not. Then again, whether or not these two things are equal is the bit where you decide on whether you want to take a guess which has its risks or hire an engineer with legal experience to sort the problem out for you.

            Comment


            • Originally posted by nanonyme View Post
              I'm not a lawyer but I'd suspect loaning ideas from there is a just fine without any license concerns but loaning code not. Then again, whether or not these two things are equal is the bit where you decide on whether you want to take a guess which has its risks or hire an engineer with legal experience to sort the problem out for you.
              Well lets just say that I am using some of their bit parsing. I thought it would be easier just using their code, instead of me having to re-invent the wheel. It is basically only code such as:

              bitstream_parameter = bitstream[0] << shift

              etc.

              Comment


              • An incorrect test which is supposed to fail but doesn't crash a GPU or Mesa means that Mesa/GPU passes and handles the incorrectness correctly. :P

                Concerning the licensing issues, you can't copy a GPL code and commit it in Mesa.

                Comment


                • FYI, I submitted the patches here:

                  http://lists.freedesktop.org/archive...st/001838.html

                  just in case anyone wants to re-test with r300g, since I only tested them on nvidia.

                  Comment


                  • Originally posted by rohcQaH View Post
                    FYI, I submitted the patches here:

                    http://lists.freedesktop.org/archive...st/001838.html

                    just in case anyone wants to re-test with r300g, since I only tested them on nvidia.
                    Tried it out with r300g, it's working fine!

                    Comment


                    • Originally posted by rohcQaH View Post
                      Looks rather good to me. I'm still suspicious about the gear not looking quite symmetric but I might just be imagining it.

                      Comment


                      • Dare I suggest having two geartrains?

                        Originally posted by rohcQaH View Post
                        So.. it's a test-case. Fixing everything may make it beautiful, but not necessarily more useful as a test-case. Does this program have a certain purpose to test specific features that must continue to be used?
                        Fix one geartrain to render beautifully, and keep the other broken one around to test that the driver doesn't crash or lock up when abused?

                        Comment


                        • Originally posted by marek View Post
                          An incorrect test which is supposed to fail but doesn't crash a GPU or Mesa means that Mesa/GPU passes and handles the incorrectness correctly. :P

                          Concerning the licensing issues, you can't copy a GPL code and commit it in Mesa.
                          Maybe if he simply states in a friendly e-mail what he likes to do (contribute to Mesa) and just ask if they could perhaps release a tiny segment of that code under the mit lisence to him they will just do it? No is what you have, a yes is what you might get.

                          Comment


                          • http://cgit.freedesktop.org/mesa/demos/
                            Brian Paul committed the patches. Now there's source from me in mesa. Ego++

                            Hopefully it'll prevent a bit of confusion for driver developers in the future.

                            Comment


                            • Originally posted by rohcQaH View Post
                              http://cgit.freedesktop.org/mesa/demos/
                              Brian Paul committed the patches. Now there's source from me in mesa. Ego++

                              Hopefully it'll prevent a bit of confusion for driver developers in the future.
                              Yeah, now there's no weird white stuff anyway. I still find it a bit confusing that the lower part of the wheel you fixed isn't similar to the upper part of the wheel but guess I can live with it.

                              Comment


                              • Originally posted by marek View Post
                                Concerning the licensing issues, you can't copy a GPL code and commit it in Mesa.
                                Well what about getting ideas from it then?

                                Comment

                                Working...
                                X