Hi all,
So I've been trying to figure out how to get Pd to run with Alsa 0.9x. I have to report that so far it's a failure.
Pd 0.33 "test" versions work with Alsa 0.9 Beta 3. However, I can't compile 0.9 Beta 3 with 2.4 kernels because Alsa 0.9B3 defines a timer_t that conflicts with the Linus 2.4 definition of the same name. You can squeak by by doing without MIDI and many offensive sound cards. If you do this you can just barely get Alsa 0.9B3 to compile in RedHat 7.1 (for example) and its 2.4 kernel. Pd 0.33 test5 (and I hope test6) then work. (You do have to know exactly what blocksize and "frags" parameters to feed it... see earlier mail from me on this list.)
I've just been entertaining myself with trying to get Pd to run with Alsa 0.9Beta 4. I had to upgrade to RedHat 7.1 because 0.9Beta4 seems not to work with 2.2 kernels. And now I find that Pd 0.33test6 (the latest) doesn't work at all with Alsa except in OSS emulation. OSS emulation seems to work OK at least.
If any of you have tried test versions of Pd 0.33 with Alsa 0.9B4 and founf out how to get it to work I'd sure like to know how you do it. In the meantime I'll continue trying to find out what my problems are.
cheers Miller
Burning the midnight oil as well, I see. :)
I have heard others have had problems with ALSA 0.9beta4 and the rawmidi code. I was advised to give up and move on to the CVS bleeding edge. Given the number of horrific bugs I've been turning up in the ALSA mtpav driver, switching to the CVS version is attractive for me.
How does one tell if Pd is using the ALSA interface or not? I got pd test 5 to compile with the CVS version of a few days ago, but I can't tell if it's using /dev/midi or something more fun when it actually runs. -alsa no longer seems to be a command line option...
Are you using the ALSA sequencer? Has the API for that stabilized?
(jfm3)
Miller Puckette wrote:
Hi all,
So I've been trying to figure out how to get Pd to run with Alsa 0.9x. I have to report that so far it's a failure.
Pd 0.33 "test" versions work with Alsa 0.9 Beta 3. However, I can't compile 0.9 Beta 3 with 2.4 kernels because Alsa 0.9B3 defines a timer_t that conflicts with the Linus 2.4 definition of the same name. You can squeak by by doing without MIDI and many offensive sound cards. If you do this you can just barely get Alsa 0.9B3 to compile in RedHat 7.1 (for example) and its 2.4 kernel. Pd 0.33 test5 (and I hope test6) then work. (You do have to know exactly what blocksize and "frags" parameters to feed it... see earlier mail from me on this list.)
I've just been entertaining myself with trying to get Pd to run with Alsa 0.9Beta 4. I had to upgrade to RedHat 7.1 because 0.9Beta4 seems not to work with 2.2 kernels. And now I find that Pd 0.33test6 (the latest) doesn't work at all with Alsa except in OSS emulation. OSS emulation seems to work OK at least.
If any of you have tried test versions of Pd 0.33 with Alsa 0.9B4 and founf out how to get it to work I'd sure like to know how you do it. In the meantime I'll continue trying to find out what my problems are.
cheers Miller
Hi,
A couple of things:
- my externals for granular sampling of an audio signal (grantab) are
now at: www.music.gla.ac.uk/nick/pd (compiled for Linux and WinNT).
- I couldn't get sox (Lin) or CoolEditPro (Win) to recognise any of the
higher-res sample formats put out by Soundfiler (floating-point Next, or 24-bit integers). So, I send a 'write' message to the table I want and use a little conversion program (ttof) to convert into fp binary (which I can open as raw PCM). Nasty workaround, but ttof is at the above URL also if anyone needs it.
If anyone has time to do an investigation of the high-res capability of Soundfiler....
cheers,
Nick
Dr Nick Fells, Music Department, University of Glasgow, Tel: +44 (0) 141-330 4096 Glasgow G12 8QQ Fax: +44 (0) 141-330 3518 UK. http://www.music.gla.ac.uk/~nick
So to summarize your and Winfried's messages, I can either get a newer ALSA _OR_ upgrade to kernel 2.4.5. I wonder what will happen if I do both...
As to the -alsa command line option being gone, try: make clean ./configure --enable-alsa make
(or --enable-old-alsa for 0.5x).
Pd uses only OSS midi; the "-alsa" flag affects only the audio code, not the MIDI code.
cheers Miller
On Mon, Jun 11, 2001 at 01:32:43AM -0400, jfm3 wrote:
Burning the midnight oil as well, I see. :)
I have heard others have had problems with ALSA 0.9beta4 and the rawmidi code. I was advised to give up and move on to the CVS bleeding edge. Given the number of horrific bugs I've been turning up in the ALSA mtpav driver, switching to the CVS version is attractive for me.
How does one tell if Pd is using the ALSA interface or not? I got pd test 5 to compile with the CVS version of a few days ago, but I can't tell if it's using /dev/midi or something more fun when it actually runs. -alsa no longer seems to be a command line option...
Are you using the ALSA sequencer? Has the API for that stabilized?
(jfm3)
Miller Puckette wrote:
Hi all,
So I've been trying to figure out how to get Pd to run with Alsa 0.9x. I have to report that so far it's a failure.
Pd 0.33 "test" versions work with Alsa 0.9 Beta 3. However, I can't compile 0.9 Beta 3 with 2.4 kernels because Alsa 0.9B3 defines a timer_t that conflicts with the Linus 2.4 definition of the same name. You can squeak by by doing without MIDI and many offensive sound cards. If you do this you can just barely get Alsa 0.9B3 to compile in RedHat 7.1 (for example) and its 2.4 kernel. Pd 0.33 test5 (and I hope test6) then work. (You do have to know exactly what blocksize and "frags" parameters to feed it... see earlier mail from me on this list.)
I've just been entertaining myself with trying to get Pd to run with Alsa 0.9Beta 4. I had to upgrade to RedHat 7.1 because 0.9Beta4 seems not to work with 2.2 kernels. And now I find that Pd 0.33test6 (the latest) doesn't work at all with Alsa except in OSS emulation. OSS emulation seems to work OK at least.
If any of you have tried test versions of Pd 0.33 with Alsa 0.9B4 and founf out how to get it to work I'd sure like to know how you do it. In the meantime I'll continue trying to find out what my problems are.
cheers Miller
Hello,
I've just been entertaining myself with trying to get Pd to run with Alsa 0.9Beta 4. I had to upgrade to RedHat 7.1 because 0.9Beta4 seems not to work with 2.2 kernels. And now I find that Pd 0.33test6 (the latest) doesn't work at all with Alsa except in OSS emulation. OSS emulation seems to work OK at least.
If any of you have tried test versions of Pd 0.33 with Alsa 0.9B4 and founf out how to get it to work I'd sure like to know how you do it. In the meantime I'll continue trying to find out what my problems are.
If got PD to work with ALSA 0.9Beta4 and 2.4.5 (but not 2.4.4 some socket open troubles) and the snd-card-rme96 (RME96/8 PST) Soundcard and MIDI with PCI128 (snd-card-ens1370) and snd-seq-oss. But I failed with rme9652, since all the anounced features (which are actually drawbacks in performance) like dev "plug:x,x", "share:x,x" or "multi:x,x" I cant get to work, even hacking on pd code (doing my own testing programm).
I got it to work with the onboard CMIPCI-Card, but using MIDI of this Card frezzes the computer, so I think some of the drivers have MIDI broken and I worked a little better (only MIDI-out) if I turn the APIC (and IO-APIC)-feature in kernel Config on (not SMP).
Also the philosophy in ALSA is that using "mmap -interface" feature turns out that you have to use a mmap-pointer update routine which in facts makes a copy of sounddata to the real mmap-memory, which in fact undermines the idea of mmap and doing the data-conversion (flot to int32) ourselves. So there I see no point to use mmap in ALSA, since it does the same: coping sound-data to the device ;-(. So I think best is using Guenthers driver for Hammerfall (which also provides an stereo device for eg. snd Soundeditor and so on).
mfg winfried
On Mon, 11 Jun 2001, Winfried Ritsch wrote:
Hello,
I've just been entertaining myself with trying to get Pd to run with Alsa 0.9Beta 4. I had to upgrade to RedHat 7.1 because 0.9Beta4 seems not to work with 2.2 kernels. And now I find that Pd 0.33test6 (the latest) doesn't work at all with Alsa except in OSS emulation. OSS emulation seems to work OK at least.
If any of you have tried test versions of Pd 0.33 with Alsa 0.9B4 and founf out how to get it to work I'd sure like to know how you do it. In the meantime I'll continue trying to find out what my problems are.
If got PD to work with ALSA 0.9Beta4 and 2.4.5 (but not 2.4.4 some socket open troubles) and the snd-card-rme96 (RME96/8 PST) Soundcard and MIDI with PCI128 (snd-card-ens1370) and snd-seq-oss. But I failed with rme9652, since all the anounced features (which are actually drawbacks in performance) like dev "plug:x,x", "share:x,x" or "multi:x,x" I cant get to work, even hacking on pd code (doing my own testing programm).
I have had similar problems and haven't had time to really look into things in depth.
I got it to work with the onboard CMIPCI-Card, but using MIDI of this Card frezzes the computer, so I think some of the drivers have MIDI broken and I worked a little better (only MIDI-out) if I turn the APIC (and IO-APIC)-feature in kernel Config on (not SMP).
Also the philosophy in ALSA is that using "mmap -interface" feature turns out that you have to use a mmap-pointer update routine which in facts makes a copy of sounddata to the real mmap-memory, which in fact undermines the idea of mmap and doing the data-conversion (flot to int32) ourselves. So there I see no point to use mmap in ALSA, since it does the same: coping sound-data to the device ;-(. So I think best is using Guenthers driver for Hammerfall (which also provides an stereo device for eg. snd Soundeditor and so on).
My understanding is that the mmap-pointer update routine only does a copy when the card can't do mmap for some reason (that is, it fakes mmap). The 'advertisement' for alsa mmap is that it is real mmap. If you are certain about this, then it yet another reason alsa is a bigger pain in the butt than it is worth. I think I am going to try Guenter's driver . . .
Karl
mfg winfried
| Karl W. MacMillan | | Computer Music Department | | Peabody Institute of the Johns Hopkins University | | karlmac@peabody.jhu.edu | | mambo.peabody.jhu.edu/~karlmac |
-
jfm3 -
Karl MacMillan -
Miller Puckette -
Nick Fells -
Winfried Ritsch