[Cs-dev] release timescale
| Date | 2008-01-30 22:19 |
| From | Victor.Lazzarini@nuim.ie |
| Subject | [Cs-dev] release timescale |
| Attachments | None None None None |
| Date | 2008-01-31 07:56 |
| From | Anthony Kozar |
| Subject | Re: [Cs-dev] release timescale |
Victor.Lazzarini@nuim.ie wrote on 1/30/08 5:19 PM: > I am just wondering now if we have a better idea of > the schedule for the release of 5.08? In particular: > - When is the final tagging taking place? I don't know -- is there still time to submit a couple of corrections to the image opcodes? I got them to compile with a couple of small changes. But, they are using fopen() instead csound->FileOpen2(). This ought to be changed (and a new filetype constant in csoundCore.h for PNG files would be appropriate). > - When will the manual be ready to go? I am finishing up my new opcode entries right now. I will commit them along with the other updates I have all together later today. > - What will be the beta-release period? > - How should we name the beta packages? > Tagging: Sunday (3) > Beta-packages: Monday-Tuesday (4-5) > Final packages: the following Monday-Tuesday (11-12) > Official release announcement: Wed 13/02/08 > > That will give us plenty of time to get things right. It usually takes me about six hours to put together my release packages since significant portions of the process are not automated. Given that only one person has ever contacted me about problems with Csound 5 on MacOS 9, I wonder about the usefulness of a "beta" release on that platform. There are some significant new opcodes though this time that I have not yet tested at all. What do you all think? Anthony ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2008-01-31 10:13 |
| From | Cesare Marilungo |
| Subject | Re: [Cs-dev] release timescale |
Anthony Kozar wrote: > Victor.Lazzarini@nuim.ie wrote on 1/30/08 5:19 PM: > > >> I am just wondering now if we have a better idea of >> the schedule for the release of 5.08? In particular: >> - When is the final tagging taking place? >> > > I don't know -- is there still time to submit a couple of corrections to the > image opcodes? I got them to compile with a couple of small changes. But, > they are using fopen() instead csound->FileOpen2(). This ought to be > changed (and a new filetype constant in csoundCore.h for PNG files would be > appropriate). > Anthony, sorry I didn't know I should have used csound->FileOpen2(), but what other changes have you made? As soon as they were included in cvs I checked and recompiled everything and they worked. -c. > >> - When will the manual be ready to go? >> > > I am finishing up my new opcode entries right now. I will commit them along > with the other updates I have all together later today. > > >> - What will be the beta-release period? >> - How should we name the beta packages? >> > > >> Tagging: Sunday (3) >> Beta-packages: Monday-Tuesday (4-5) >> Final packages: the following Monday-Tuesday (11-12) >> Official release announcement: Wed 13/02/08 >> >> That will give us plenty of time to get things right. >> > > It usually takes me about six hours to put together my release packages > since significant portions of the process are not automated. Given that > only one person has ever contacted me about problems with Csound 5 on MacOS > 9, I wonder about the usefulness of a "beta" release on that platform. > There are some significant new opcodes though this time that I have not yet > tested at all. What do you all think? > > Anthony > > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > > > -- www.cesaremarilungo.com ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2008-01-31 20:14 |
| From | Anthony Kozar |
| Subject | Re: [Cs-dev] release timescale |
Cesare Marilungo wrote on 1/31/08 5:13 AM: > Anthony, sorry I didn't know I should have used csound->FileOpen2(), but Not a problem. FileOpen2() provides a number of important advantages over fopen() though. With it, you can set a search path for finding files such as "SFDIR;SSDIR" -- otherwise files have to be in the current directory unless a full path is given. (And the current directory doesn't make sense a lot of times for GUI apps). Csound will also convert pathnames to the current platform's conventions. The other advantage is that Csound knows you have opened a file and ensures that it is closed at the end of the performance. Csound also notifies the host about it. On MacOS, the front end uses this information to provide a better user experience. Csound has other functions for finding and opening files too. For loading large files whose data is only read, not changed, there is ldmemfile2(). This function ensures that even if multiple instruments request loading the same file, that it is only loaded into memory once. > what other changes have you made? As soon as they were included in cvs I > checked and recompiled everything and they worked. I had to change #include |
| Date | 2008-01-31 20:37 |
| From | Cesare Marilungo |
| Subject | Re: [Cs-dev] release timescale |
Thanks for the clarifications, Anthony. -c. Anthony Kozar wrote: > Cesare Marilungo wrote on 1/31/08 5:13 AM: > > >> Anthony, sorry I didn't know I should have used csound->FileOpen2(), but >> > > Not a problem. FileOpen2() provides a number of important advantages over > fopen() though. With it, you can set a search path for finding files such > as "SFDIR;SSDIR" -- otherwise files have to be in the current directory > unless a full path is given. (And the current directory doesn't make sense > a lot of times for GUI apps). Csound will also convert pathnames to the > current platform's conventions. > > The other advantage is that Csound knows you have opened a file and ensures > that it is closed at the end of the performance. Csound also notifies the > host about it. On MacOS, the front end uses this information to provide a > better user experience. > > Csound has other functions for finding and opening files too. For loading > large files whose data is only read, not changed, there is ldmemfile2(). > This function ensures that even if multiple instruments request loading the > same file, that it is only loaded into memory once. > > >> what other changes have you made? As soon as they were included in cvs I >> checked and recompiled everything and they worked. >> > > I had to change > > #include |
| Date | 2008-01-31 22:32 |
| From | Anthony Kozar |
| Subject | Re: [Cs-dev] image opcodes |
These changes are all done now. I've set imageload to search in current dir, then SSDIR and SFDIR. imagesave still only saves to current dir. These could be changed of course. I tested with example from the manual and images load and save fine on MacOS 9. This is cool :) The example copied an image and with the one I tried, I ended up with one row of pixels at the bottom with random colors. I am not sure if this is the CSD's fault or a bug in one of the opcodes ... Anthony Anthony Kozar wrote on 1/31/08 3:14 PM: > I will commit these simple changes myself today if the sources have not been > tagged yet. I can do the FileOpen2() change too if that is OK with everyone. ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2008-01-31 22:52 |
| From | Cesare Marilungo |
| Subject | Re: [Cs-dev] image opcodes |
Anthony Kozar wrote: > These changes are all done now. I've set imageload to search in current > dir, then SSDIR and SFDIR. imagesave still only saves to current dir. > These could be changed of course. > > I tested with example from the manual and images load and save fine on MacOS > 9. This is cool :) The example copied an image and with the one I tried, I > ended up with one row of pixels at the bottom with random colors. I am not > sure if this is the CSD's fault or a bug in one of the opcodes ... > Can you please send me the image? My guess is that you haven't changed the size dependent parts of the code. Please try with the .csd below, in which I have parametrized the image size: |
| Date | 2008-01-31 23:52 |
| From | Anthony Kozar |
| Subject | Re: [Cs-dev] image opcodes |
| Attachments | image.png imageout.png None None |
Cesare Marilungo wrote on 1/31/08 5:52 PM: > Can you please send me the image? I've attached the input and the output. I'm thinking that maybe the "kx_ linseg 0, p3, 1" statement never quite reaches 1.0? > My guess is that you haven't changed the size dependent parts of the > code. Please try with the .csd below, in which I have parametrized the > image size: The example in the manual is almost identical to the one you sent with three exceptions, but the latter still does not work. > ky_ = kndx/(giimageh - 1) This is ky_ = kndx/(giimagew - 1) in the manual example. > loop_lt kndx, 1, (giimageh - 1), myloop This is loop_lt kndx, 1, (giimagew - 1), myloop in the manual example. > i3 3 1 This is i2 3 1 in the manual example. Also, I noticed that the get/set opcodes can crash if the x or y value is out of bounds. It would be better in my opinion if they didn't. (It crashes the host application too ...) Anthony |
| Date | 2008-01-31 23:58 |
| From | Cesare Marilungo |
| Subject | Re: [Cs-dev] image opcodes |
Yes, while tweaking with the example I've noticed that it can crash. Big time! :-) I'm fixing it. Next I'll see what doesn't work with this example, but consider that using a float range from 0 to 1 doesn't allow a precise indexing of pixels. At best it would reread or rewrite pixels more than once. Of course these opcodes wheren't meant to do image processing. Stay tuned. -c. Anthony Kozar wrote: > Cesare Marilungo wrote on 1/31/08 5:52 PM: > > >> Can you please send me the image? >> > > I've attached the input and the output. > > I'm thinking that maybe the "kx_ linseg 0, p3, 1" statement never quite > reaches 1.0? > > >> My guess is that you haven't changed the size dependent parts of the >> code. Please try with the .csd below, in which I have parametrized the >> image size: >> > > The example in the manual is almost identical to the one you sent with three > exceptions, but the latter still does not work. > > >> ky_ = kndx/(giimageh - 1) >> > > This is ky_ = kndx/(giimagew - 1) > in the manual example. > > >> loop_lt kndx, 1, (giimageh - 1), myloop >> > > This is loop_lt kndx, 1, (giimagew - 1), myloop > in the manual example. > > >> i3 3 1 >> > > This is i2 3 1 in the manual example. > > Also, I noticed that the get/set opcodes can crash if the x or y value is > out of bounds. It would be better in my opinion if they didn't. (It > crashes the host application too ...) > > Anthony > > > ------------------------------------------------------------------------ > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > ------------------------------------------------------------------------ > > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > -- www.cesaremarilungo.com ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2008-02-01 00:37 |
| From | Cesare Marilungo |
| Subject | Re: [Cs-dev] image opcodes |
Hi Anthony, I've corrected the bounds checking. x and y should have been < (not <=) than respectively w and h. Could you please check it back from cvs and test it? Best, c-. Anthony Kozar wrote: > Cesare Marilungo wrote on 1/31/08 5:52 PM: > > >> Can you please send me the image? >> > > I've attached the input and the output. > > I'm thinking that maybe the "kx_ linseg 0, p3, 1" statement never quite > reaches 1.0? > > >> My guess is that you haven't changed the size dependent parts of the >> code. Please try with the .csd below, in which I have parametrized the >> image size: >> > > The example in the manual is almost identical to the one you sent with three > exceptions, but the latter still does not work. > > >> ky_ = kndx/(giimageh - 1) >> > > This is ky_ = kndx/(giimagew - 1) > in the manual example. > > >> loop_lt kndx, 1, (giimageh - 1), myloop >> > > This is loop_lt kndx, 1, (giimagew - 1), myloop > in the manual example. > > >> i3 3 1 >> > > This is i2 3 1 in the manual example. > > Also, I noticed that the get/set opcodes can crash if the x or y value is > out of bounds. It would be better in my opinion if they didn't. (It > crashes the host application too ...) > > Anthony > > > ------------------------------------------------------------------------ > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > ------------------------------------------------------------------------ > > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > -- www.cesaremarilungo.com ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2008-02-01 00:56 |
| From | Cesare Marilungo |
| Subject | Re: [Cs-dev] image opcodes |
And with this .csd example below the copied image should be right. I just changed the loop increment step from 1 to 0.5. We should update this in the manual. Best, -c. |
| Date | 2008-02-01 04:37 |
| From | Anthony Kozar |
| Subject | Re: [Cs-dev] image opcodes |
Excellent! This newest example appears to work perfectly. (Maybe the opcode should do the rounding of the pixel position if it doesn't?) It did not crash either when I forced the y position out of bounds. No warning either -- but a warning every k-loop could be exhausting -- should a note deactivate itself or continue when such an error occurs? Great work! Thanks, Cesare. Anthony Cesare Marilungo wrote on 1/31/08 7:56 PM: > And with this .csd example below the copied image should be right. I > just changed the loop increment step from 1 to 0.5. > > We should update this in the manual. ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2008-02-01 10:48 |
| From | Cesare Marilungo |
| Subject | Re: [Cs-dev] image opcodes |
Anthony Kozar wrote: > Excellent! This newest example appears to work perfectly. (Maybe the opcode > should do the rounding of the pixel position if it doesn't?) > The opcode multiplies the 0..1 float values (for x and y) for the image size (width and height respectively) and casts to int. So by decreasing the loop increment step in the example I am sure he gets all the rows (even though it will process the same row more than once). For musical purposes one doesn't necessarily need to read all the pixels, and a low pass filter can be applied to smooth the resulting signal. For the next version I might try to add some kind of interpolation. > It did not crash either when I forced the y position out of bounds. No > warning either -- but a warning every k-loop could be exhausting -- should a > note deactivate itself or continue when such an error occurs? > > I think it should just continue and return 0 for r,g and b (a black pixel) like it does now, or alternatively return the value at the margin. > Great work! Thanks, Cesare. > Thank you. But there's still room for improvement, for sure. -c. > Anthony > > Cesare Marilungo wrote on 1/31/08 7:56 PM: > > >> And with this .csd example below the copied image should be right. I >> just changed the loop increment step from 1 to 0.5. >> >> We should update this in the manual. >> > > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > > > -- www.cesaremarilungo.com ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |