hello,
i just did a initial commit of share-mem, a lib dedicated to deal with shared memory.
background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability.
Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples.
Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows.
everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development.
Cheers Cyrille
Pd-announce mailing list Pd-announce@iem.at http://lists.puredata.info/listinfo/pd-announce
Sounds interesting, where do we find it?
.hc
On Oct 31, 2012, at 7:19 PM, Cyrille Henry wrote:
hello,
i just did a initial commit of share-mem, a lib dedicated to deal with shared memory.
background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability.
Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples.
Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows.
everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development.
Cheers Cyrille
Pd-announce mailing list Pd-announce@iem.at http://lists.puredata.info/listinfo/pd-announce
Le 01/11/2012 04:04, Hans-Christoph Steiner a écrit :
Sounds interesting, where do we find it?
i commit everything in pd svn : trunk/externals/share-mem
cheers c
.hc
On Oct 31, 2012, at 7:19 PM, Cyrille Henry wrote:
hello,
i just did a initial commit of share-mem, a lib dedicated to deal with shared memory.
background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability.
Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples.
Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows.
everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development.
Cheers Cyrille
Pd-announce mailing list Pd-announce@iem.at http://lists.puredata.info/listinfo/pd-announce
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Cyrille, this is very exciting. Where can we get more details?
Phil
On 10/31/12 4:19 PM, Cyrille Henry wrote:
hello,
i just did a initial commit of share-mem, a lib dedicated to deal with shared memory.
background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability.
Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples.
Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows.
everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development.
Cheers Cyrille
Pd-announce mailing list Pd-announce@iem.at http://lists.puredata.info/listinfo/pd-announce
hello,
for now, you'll have to grab sources from svn and compile it yourself. examples should provide a quick overview of the usability. cheers c
Le 01/11/2012 05:40, Phil Stone a écrit :
Cyrille, this is very exciting. Where can we get more details?
Phil
On 10/31/12 4:19 PM, Cyrille Henry wrote:
hello,
i just did a initial commit of share-mem, a lib dedicated to deal with shared memory.
background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability.
Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples.
Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows.
everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development.
Cheers Cyrille
Pd-announce mailing list Pd-announce@iem.at http://lists.puredata.info/listinfo/pd-announce
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
btw : the object is still in development and may improve these days specially concerning the message syntax. we may provide osx (and ubuntu) binaries after the testing period. best, n
Le 01/11/12 10:24, Cyrille Henry a écrit :
hello,
for now, you'll have to grab sources from svn and compile it yourself. examples should provide a quick overview of the usability. cheers c
Le 01/11/2012 05:40, Phil Stone a écrit :
Cyrille, this is very exciting. Where can we get more details?
Phil
On 10/31/12 4:19 PM, Cyrille Henry wrote:
hello,
i just did a initial commit of share-mem, a lib dedicated to deal with shared memory.
background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability.
Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples.
Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows.
everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development.
Cheers Cyrille
Pd-announce mailing list Pd-announce@iem.at http://lists.puredata.info/listinfo/pd-announce
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
You can get binaries from any library in pure-data SVN based on the Library Template from the Jenkins build server. There will soon be a Windows builder there too. For example, here's a Mac OS X 32-bit i386 binary for shmem.pd_darwin:
https://macosx105-i386.pdlab.puredata.info/job/template-libraries/label=maco...
You can browse available platforms here: https://macosx105-i386.pdlab.puredata.info/job/template-libraries/
Also, there is a shmem.pd_linux checked into SVN, that should be removed since it can be generated by typing 'make'.
.hc
On Nov 1, 2012, at 7:15 AM, Nicolas Montgermont wrote:
btw : the object is still in development and may improve these days specially concerning the message syntax. we may provide osx (and ubuntu) binaries after the testing period. best, n
Le 01/11/12 10:24, Cyrille Henry a écrit :
hello,
for now, you'll have to grab sources from svn and compile it yourself. examples should provide a quick overview of the usability. cheers c
Le 01/11/2012 05:40, Phil Stone a écrit :
Cyrille, this is very exciting. Where can we get more details?
Phil
On 10/31/12 4:19 PM, Cyrille Henry wrote:
hello,
i just did a initial commit of share-mem, a lib dedicated to deal with shared memory.
background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability.
Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples.
Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows.
everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development.
Cheers Cyrille
Pd-announce mailing list Pd-announce@iem.at http://lists.puredata.info/listinfo/pd-announce
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Le 01/11/2012 16:30, Hans-Christoph Steiner a écrit :
You can get binaries from any library in pure-data SVN based on the Library Template from the Jenkins build server. There will soon be a Windows builder there too. For example, here's a Mac OS X 32-bit i386 binary for shmem.pd_darwin:
https://macosx105-i386.pdlab.puredata.info/job/template-libraries/label=maco...
You can browse available platforms here: https://macosx105-i386.pdlab.puredata.info/job/template-libraries/
Also, there is a shmem.pd_linux checked into SVN, that should be removed since it can be generated by typing 'make'.
oups, my mistake. i removed it. cheers c
.hc
On Nov 1, 2012, at 7:15 AM, Nicolas Montgermont wrote:
btw : the object is still in development and may improve these days specially concerning the message syntax. we may provide osx (and ubuntu) binaries after the testing period. best, n
Le 01/11/12 10:24, Cyrille Henry a écrit :
hello,
for now, you'll have to grab sources from svn and compile it yourself. examples should provide a quick overview of the usability. cheers c
Le 01/11/2012 05:40, Phil Stone a écrit :
Cyrille, this is very exciting. Where can we get more details?
Phil
On 10/31/12 4:19 PM, Cyrille Henry wrote:
hello,
i just did a initial commit of share-mem, a lib dedicated to deal with shared memory.
background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability.
Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples.
Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows.
everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development.
Cheers Cyrille
Pd-announce mailing list Pd-announce@iem.at http://lists.puredata.info/listinfo/pd-announce
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
nice work thanks guys,
maybe you can add a [declare -path ..] to all examples patches so that they will work just after the make
++ a
yes, thanks! c
Le 02/11/2012 12:34, Antoine Villeret a écrit :
nice work thanks guys,
maybe you can add a [declare -path ..] to all examples patches so that they will work just after the make
++ a
-- do it yourself http://antoine.villeret.free.fr http://drii.ensad.fr -- Google lit ce mail... si vous refusez cela, utilisez l'adresse antoine.villeret [at] free.fr http://free.fr pour me contacter
2012/11/1 Cyrille Henry <ch@chnry.net mailto:ch@chnry.net>
Le 01/11/2012 16:30, Hans-Christoph Steiner a écrit : You can get binaries from any library in pure-data SVN based on the Library Template from the Jenkins build server. There will soon be a Windows builder there too. For example, here's a Mac OS X 32-bit i386 binary for shmem.pd_darwin: https://macosx105-i386.pdlab.__puredata.info/job/template-__libraries/label=macosx105-__i386/ws/share_mem/shmem.pd___darwin <https://macosx105-i386.pdlab.puredata.info/job/template-libraries/label=macosx105-i386/ws/share_mem/shmem.pd_darwin> You can browse available platforms here: https://macosx105-i386.pdlab.__puredata.info/job/template-__libraries/ <https://macosx105-i386.pdlab.puredata.info/job/template-libraries/> Also, there is a shmem.pd_linux checked into SVN, that should be removed since it can be generated by typing 'make'. oups, my mistake. i removed it. cheers c .hc On Nov 1, 2012, at 7:15 AM, Nicolas Montgermont wrote: btw : the object is still in development and may improve these days specially concerning the message syntax. we may provide osx (and ubuntu) binaries after the testing period. best, n Le 01/11/12 10:24, Cyrille Henry a écrit : hello, for now, you'll have to grab sources from svn and compile it yourself. examples should provide a quick overview of the usability. cheers c Le 01/11/2012 05:40, Phil Stone a écrit : Cyrille, this is very exciting. Where can we get more details? Phil On 10/31/12 4:19 PM, Cyrille Henry wrote: hello, i just did a initial commit of share-mem, a lib dedicated to deal with shared memory. background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability. ----- Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples. Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows. everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development. Cheers Cyrille _________________________________________________ Pd-announce mailing list Pd-announce@iem.at <mailto:Pd-announce@iem.at> http://lists.puredata.info/__listinfo/pd-announce <http://lists.puredata.info/listinfo/pd-announce> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> -- http://nim.on.free.fr _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list>
The downside of using [declare -path ..] is that if someone edits an example and saves it somewhere else, it'll no longer work. An alternative is to use [declare -lib share-mem], then it'll work wherever as long as share-mem is installed. The downside there is that the library needs to be in the standard install location or the path in order for that to work. For development, I often put a symlink in ~/pd-externals to the folder in SVN, then it works everywhere.
.hc
On Nov 2, 2012, at 7:48 AM, Cyrille Henry wrote:
yes, thanks! c
Le 02/11/2012 12:34, Antoine Villeret a écrit :
nice work thanks guys,
maybe you can add a [declare -path ..] to all examples patches so that they will work just after the make
++ a
-- do it yourself http://antoine.villeret.free.fr http://drii.ensad.fr -- Google lit ce mail... si vous refusez cela, utilisez l'adresse antoine.villeret [at] free.fr http://free.fr pour me contacter
2012/11/1 Cyrille Henry <ch@chnry.net mailto:ch@chnry.net>
Le 01/11/2012 16:30, Hans-Christoph Steiner a écrit :
You can get binaries from any library in pure-data SVN based on the Library Template from the Jenkins build server. There will soon be a Windows builder there too. For example, here's a Mac OS X 32-bit i386 binary for shmem.pd_darwin: https://macosx105-i386.pdlab.__puredata.info/job/template-__libraries/label=macosx105-__i386/ws/share_mem/shmem.pd___darwin <https://macosx105-i386.pdlab.puredata.info/job/template-libraries/label=macosx105-i386/ws/share_mem/shmem.pd_darwin> You can browse available platforms here: https://macosx105-i386.pdlab.__puredata.info/job/template-__libraries/ <https://macosx105-i386.pdlab.puredata.info/job/template-libraries/> Also, there is a shmem.pd_linux checked into SVN, that should be removed since it can be generated by typing 'make'.oups, my mistake. i removed it. cheers
c
.hc On Nov 1, 2012, at 7:15 AM, Nicolas Montgermont wrote: btw : the object is still in development and may improve these days specially concerning the message syntax. we may provide osx (and ubuntu) binaries after the testing period. best, n Le 01/11/12 10:24, Cyrille Henry a écrit : hello, for now, you'll have to grab sources from svn and compile it yourself. examples should provide a quick overview of the usability. cheers c Le 01/11/2012 05:40, Phil Stone a écrit : Cyrille, this is very exciting. Where can we get more details? Phil On 10/31/12 4:19 PM, Cyrille Henry wrote: hello, i just did a initial commit of share-mem, a lib dedicated to deal with shared memory. background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability. ----- Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples. Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows. everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development. Cheers Cyrille _________________________________________________ Pd-announce mailing list Pd-announce@iem.at <mailto:Pd-announce@iem.at> http://lists.puredata.info/__listinfo/pd-announce <http://lists.puredata.info/listinfo/pd-announce> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> -- http://nim.on.free.fr _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list>
Pd-list@iem.at mailto:Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
However, the "declare -path" won't make it _not_ work and will fix the example if a 'user' downloads it without changing their Pd installation (as someone like me would normally do.) So I'd recommend the "-path" option as the easier way to get started up. Once correctly "installed", ideally, it shouldn't need any declarations at all.
cheers M
On Fri, Nov 02, 2012 at 11:03:36AM -0400, Hans-Christoph Steiner wrote:
The downside of using [declare -path ..] is that if someone edits an example and saves it somewhere else, it'll no longer work. An alternative is to use [declare -lib share-mem], then it'll work wherever as long as share-mem is installed. The downside there is that the library needs to be in the standard install location or the path in order for that to work. For development, I often put a symlink in ~/pd-externals to the folder in SVN, then it works everywhere.
.hc
On Nov 2, 2012, at 7:48 AM, Cyrille Henry wrote:
yes, thanks! c
Le 02/11/2012 12:34, Antoine Villeret a écrit :
nice work thanks guys,
maybe you can add a [declare -path ..] to all examples patches so that they will work just after the make
++ a
-- do it yourself http://antoine.villeret.free.fr http://drii.ensad.fr -- Google lit ce mail... si vous refusez cela, utilisez l'adresse antoine.villeret [at] free.fr http://free.fr pour me contacter
2012/11/1 Cyrille Henry <ch@chnry.net mailto:ch@chnry.net>
Le 01/11/2012 16:30, Hans-Christoph Steiner a écrit :
You can get binaries from any library in pure-data SVN based on the Library Template from the Jenkins build server. There will soon be a Windows builder there too. For example, here's a Mac OS X 32-bit i386 binary for shmem.pd_darwin: https://macosx105-i386.pdlab.__puredata.info/job/template-__libraries/label=macosx105-__i386/ws/share_mem/shmem.pd___darwin <https://macosx105-i386.pdlab.puredata.info/job/template-libraries/label=macosx105-i386/ws/share_mem/shmem.pd_darwin> You can browse available platforms here: https://macosx105-i386.pdlab.__puredata.info/job/template-__libraries/ <https://macosx105-i386.pdlab.puredata.info/job/template-libraries/> Also, there is a shmem.pd_linux checked into SVN, that should be removed since it can be generated by typing 'make'.oups, my mistake. i removed it. cheers
c
.hc On Nov 1, 2012, at 7:15 AM, Nicolas Montgermont wrote: btw : the object is still in development and may improve these days specially concerning the message syntax. we may provide osx (and ubuntu) binaries after the testing period. best, n Le 01/11/12 10:24, Cyrille Henry a écrit : hello, for now, you'll have to grab sources from svn and compile it yourself. examples should provide a quick overview of the usability. cheers c Le 01/11/2012 05:40, Phil Stone a écrit : Cyrille, this is very exciting. Where can we get more details? Phil On 10/31/12 4:19 PM, Cyrille Henry wrote: hello, i just did a initial commit of share-mem, a lib dedicated to deal with shared memory. background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability. ----- Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples. Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows. everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development. Cheers Cyrille _________________________________________________ Pd-announce mailing list Pd-announce@iem.at <mailto:Pd-announce@iem.at> http://lists.puredata.info/__listinfo/pd-announce <http://lists.puredata.info/listinfo/pd-announce> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> -- http://nim.on.free.fr _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list>
Pd-list@iem.at mailto:Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
No declaration at all will take us to the bad old days when there was no way to resolve which [prepend], [counter], [scale], etc. that a given patch needed except to guess. Or if you find the patch on the net, the [declare -lib] will tell you that you need to get the 'share-mem' library in order for the patch to work. Otherwise, you have to guess where [shmem] came from.
.hc
On Nov 2, 2012, at 11:08 AM, Miller Puckette wrote:
However, the "declare -path" won't make it _not_ work and will fix the example if a 'user' downloads it without changing their Pd installation (as someone like me would normally do.) So I'd recommend the "-path" option as the easier way to get started up. Once correctly "installed", ideally, it shouldn't need any declarations at all.
cheers M
On Fri, Nov 02, 2012 at 11:03:36AM -0400, Hans-Christoph Steiner wrote:
The downside of using [declare -path ..] is that if someone edits an example and saves it somewhere else, it'll no longer work. An alternative is to use [declare -lib share-mem], then it'll work wherever as long as share-mem is installed. The downside there is that the library needs to be in the standard install location or the path in order for that to work. For development, I often put a symlink in ~/pd-externals to the folder in SVN, then it works everywhere.
.hc
On Nov 2, 2012, at 7:48 AM, Cyrille Henry wrote:
yes, thanks! c
Le 02/11/2012 12:34, Antoine Villeret a écrit :
nice work thanks guys,
maybe you can add a [declare -path ..] to all examples patches so that they will work just after the make
++ a
-- do it yourself http://antoine.villeret.free.fr http://drii.ensad.fr -- Google lit ce mail... si vous refusez cela, utilisez l'adresse antoine.villeret [at] free.fr http://free.fr pour me contacter
2012/11/1 Cyrille Henry <ch@chnry.net mailto:ch@chnry.net>
Le 01/11/2012 16:30, Hans-Christoph Steiner a écrit :
You can get binaries from any library in pure-data SVN based on the Library Template from the Jenkins build server. There will soon be a Windows builder there too. For example, here's a Mac OS X 32-bit i386 binary for shmem.pd_darwin: https://macosx105-i386.pdlab.__puredata.info/job/template-__libraries/label=macosx105-__i386/ws/share_mem/shmem.pd___darwin <https://macosx105-i386.pdlab.puredata.info/job/template-libraries/label=macosx105-i386/ws/share_mem/shmem.pd_darwin> You can browse available platforms here: https://macosx105-i386.pdlab.__puredata.info/job/template-__libraries/ <https://macosx105-i386.pdlab.puredata.info/job/template-libraries/> Also, there is a shmem.pd_linux checked into SVN, that should be removed since it can be generated by typing 'make'.oups, my mistake. i removed it. cheers
c
.hc On Nov 1, 2012, at 7:15 AM, Nicolas Montgermont wrote: btw : the object is still in development and may improve these days specially concerning the message syntax. we may provide osx (and ubuntu) binaries after the testing period. best, n Le 01/11/12 10:24, Cyrille Henry a écrit : hello, for now, you'll have to grab sources from svn and compile it yourself. examples should provide a quick overview of the usability. cheers c Le 01/11/2012 05:40, Phil Stone a écrit : Cyrille, this is very exciting. Where can we get more details? Phil On 10/31/12 4:19 PM, Cyrille Henry wrote: hello, i just did a initial commit of share-mem, a lib dedicated to deal with shared memory. background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability. ----- Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples. Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows. everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development. Cheers Cyrille _________________________________________________ Pd-announce mailing list Pd-announce@iem.at <mailto:Pd-announce@iem.at> http://lists.puredata.info/__listinfo/pd-announce <http://lists.puredata.info/listinfo/pd-announce> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> -- http://nim.on.free.fr _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list>
Pd-list@iem.at mailto:Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Sorry, my post was ambiguous - I meant to suggest that "declare -path ..." was a good idea for the example patches so that they'd work out of the box for people downloading shmem to temporary quarters to look at it and/or to include it in a pd-version-proof 'patchball' (if I may coin a term :)
cheers Miller
On Fri, Nov 02, 2012 at 11:40:38AM -0400, Hans-Christoph Steiner wrote:
No declaration at all will take us to the bad old days when there was no way to resolve which [prepend], [counter], [scale], etc. that a given patch needed except to guess. Or if you find the patch on the net, the [declare -lib] will tell you that you need to get the 'share-mem' library in order for the patch to work. Otherwise, you have to guess where [shmem] came from.
.hc
On Nov 2, 2012, at 11:08 AM, Miller Puckette wrote:
However, the "declare -path" won't make it _not_ work and will fix the example if a 'user' downloads it without changing their Pd installation (as someone like me would normally do.) So I'd recommend the "-path" option as the easier way to get started up. Once correctly "installed", ideally, it shouldn't need any declarations at all.
cheers M
On Fri, Nov 02, 2012 at 11:03:36AM -0400, Hans-Christoph Steiner wrote:
The downside of using [declare -path ..] is that if someone edits an example and saves it somewhere else, it'll no longer work. An alternative is to use [declare -lib share-mem], then it'll work wherever as long as share-mem is installed. The downside there is that the library needs to be in the standard install location or the path in order for that to work. For development, I often put a symlink in ~/pd-externals to the folder in SVN, then it works everywhere.
.hc
On Nov 2, 2012, at 7:48 AM, Cyrille Henry wrote:
yes, thanks! c
Le 02/11/2012 12:34, Antoine Villeret a écrit :
nice work thanks guys,
maybe you can add a [declare -path ..] to all examples patches so that they will work just after the make
++ a
-- do it yourself http://antoine.villeret.free.fr http://drii.ensad.fr -- Google lit ce mail... si vous refusez cela, utilisez l'adresse antoine.villeret [at] free.fr http://free.fr pour me contacter
2012/11/1 Cyrille Henry <ch@chnry.net mailto:ch@chnry.net>
Le 01/11/2012 16:30, Hans-Christoph Steiner a écrit :
You can get binaries from any library in pure-data SVN based on the Library Template from the Jenkins build server. There will soon be a Windows builder there too. For example, here's a Mac OS X 32-bit i386 binary for shmem.pd_darwin: https://macosx105-i386.pdlab.__puredata.info/job/template-__libraries/label=macosx105-__i386/ws/share_mem/shmem.pd___darwin <https://macosx105-i386.pdlab.puredata.info/job/template-libraries/label=macosx105-i386/ws/share_mem/shmem.pd_darwin> You can browse available platforms here: https://macosx105-i386.pdlab.__puredata.info/job/template-__libraries/ <https://macosx105-i386.pdlab.puredata.info/job/template-libraries/> Also, there is a shmem.pd_linux checked into SVN, that should be removed since it can be generated by typing 'make'.oups, my mistake. i removed it. cheers
c
.hc On Nov 1, 2012, at 7:15 AM, Nicolas Montgermont wrote: btw : the object is still in development and may improve these days specially concerning the message syntax. we may provide osx (and ubuntu) binaries after the testing period. best, n Le 01/11/12 10:24, Cyrille Henry a écrit : hello, for now, you'll have to grab sources from svn and compile it yourself. examples should provide a quick overview of the usability. cheers c Le 01/11/2012 05:40, Phil Stone a écrit : Cyrille, this is very exciting. Where can we get more details? Phil On 10/31/12 4:19 PM, Cyrille Henry wrote: hello, i just did a initial commit of share-mem, a lib dedicated to deal with shared memory. background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability. ----- Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples. Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows. everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development. Cheers Cyrille _________________________________________________ Pd-announce mailing list Pd-announce@iem.at <mailto:Pd-announce@iem.at> http://lists.puredata.info/__listinfo/pd-announce <http://lists.puredata.info/listinfo/pd-announce> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> -- http://nim.on.free.fr _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list>
Pd-list@iem.at mailto:Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
The sure fire way would be to do:
[declare -lib share-mem -path ..]
.hc
On Nov 2, 2012, at 11:47 AM, Miller Puckette wrote:
Sorry, my post was ambiguous - I meant to suggest that "declare -path ..." was a good idea for the example patches so that they'd work out of the box for people downloading shmem to temporary quarters to look at it and/or to include it in a pd-version-proof 'patchball' (if I may coin a term :)
cheers Miller
On Fri, Nov 02, 2012 at 11:40:38AM -0400, Hans-Christoph Steiner wrote:
No declaration at all will take us to the bad old days when there was no way to resolve which [prepend], [counter], [scale], etc. that a given patch needed except to guess. Or if you find the patch on the net, the [declare -lib] will tell you that you need to get the 'share-mem' library in order for the patch to work. Otherwise, you have to guess where [shmem] came from.
.hc
On Nov 2, 2012, at 11:08 AM, Miller Puckette wrote:
However, the "declare -path" won't make it _not_ work and will fix the example if a 'user' downloads it without changing their Pd installation (as someone like me would normally do.) So I'd recommend the "-path" option as the easier way to get started up. Once correctly "installed", ideally, it shouldn't need any declarations at all.
cheers M
On Fri, Nov 02, 2012 at 11:03:36AM -0400, Hans-Christoph Steiner wrote:
The downside of using [declare -path ..] is that if someone edits an example and saves it somewhere else, it'll no longer work. An alternative is to use [declare -lib share-mem], then it'll work wherever as long as share-mem is installed. The downside there is that the library needs to be in the standard install location or the path in order for that to work. For development, I often put a symlink in ~/pd-externals to the folder in SVN, then it works everywhere.
.hc
On Nov 2, 2012, at 7:48 AM, Cyrille Henry wrote:
yes, thanks! c
Le 02/11/2012 12:34, Antoine Villeret a écrit :
nice work thanks guys,
maybe you can add a [declare -path ..] to all examples patches so that they will work just after the make
++ a
-- do it yourself http://antoine.villeret.free.fr http://drii.ensad.fr -- Google lit ce mail... si vous refusez cela, utilisez l'adresse antoine.villeret [at] free.fr http://free.fr pour me contacter
2012/11/1 Cyrille Henry <ch@chnry.net mailto:ch@chnry.net>
Le 01/11/2012 16:30, Hans-Christoph Steiner a écrit :
You can get binaries from any library in pure-data SVN based on the Library Template from the Jenkins build server. There will soon be a Windows builder there too. For example, here's a Mac OS X 32-bit i386 binary for shmem.pd_darwin: https://macosx105-i386.pdlab.__puredata.info/job/template-__libraries/label=macosx105-__i386/ws/share_mem/shmem.pd___darwin <https://macosx105-i386.pdlab.puredata.info/job/template-libraries/label=macosx105-i386/ws/share_mem/shmem.pd_darwin> You can browse available platforms here: https://macosx105-i386.pdlab.__puredata.info/job/template-__libraries/ <https://macosx105-i386.pdlab.puredata.info/job/template-libraries/> Also, there is a shmem.pd_linux checked into SVN, that should be removed since it can be generated by typing 'make'.oups, my mistake. i removed it. cheers
c
.hc On Nov 1, 2012, at 7:15 AM, Nicolas Montgermont wrote: btw : the object is still in development and may improve these days specially concerning the message syntax. we may provide osx (and ubuntu) binaries after the testing period. best, n Le 01/11/12 10:24, Cyrille Henry a écrit : hello, for now, you'll have to grab sources from svn and compile it yourself. examples should provide a quick overview of the usability. cheers c Le 01/11/2012 05:40, Phil Stone a écrit : Cyrille, this is very exciting. Where can we get more details? Phil On 10/31/12 4:19 PM, Cyrille Henry wrote: hello, i just did a initial commit of share-mem, a lib dedicated to deal with shared memory. background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability. ----- Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples. Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows. everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development. Cheers Cyrille _________________________________________________ Pd-announce mailing list Pd-announce@iem.at <mailto:Pd-announce@iem.at> http://lists.puredata.info/__listinfo/pd-announce <http://lists.puredata.info/listinfo/pd-announce> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> -- http://nim.on.free.fr _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list> _________________________________________________ Pd-list@iem.at <mailto:Pd-list@iem.at> mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list <http://lists.puredata.info/listinfo/pd-list>
Pd-list@iem.at mailto:Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/__listinfo/pd-list http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
are those links dead? m.
Am 01.11.2012 um 16:30 schrieb Hans-Christoph Steiner hans@at.or.at:
You can get binaries from any library in pure-data SVN based on the Library Template from the Jenkins build server. There will soon be a Windows builder there too. For example, here's a Mac OS X 32-bit i386 binary for shmem.pd_darwin:
https://macosx105-i386.pdlab.puredata.info/job/template-libraries/label=maco...
You can browse available platforms here: https://macosx105-i386.pdlab.puredata.info/job/template-libraries/
Also, there is a shmem.pd_linux checked into SVN, that should be removed since it can be generated by typing 'make'.
.hc
On Nov 1, 2012, at 7:15 AM, Nicolas Montgermont wrote:
btw : the object is still in development and may improve these days specially concerning the message syntax. we may provide osx (and ubuntu) binaries after the testing period. best, n
Le 01/11/12 10:24, Cyrille Henry a écrit :
hello,
for now, you'll have to grab sources from svn and compile it yourself. examples should provide a quick overview of the usability. cheers c
Le 01/11/2012 05:40, Phil Stone a écrit :
Cyrille, this is very exciting. Where can we get more details?
Phil
On 10/31/12 4:19 PM, Cyrille Henry wrote:
hello,
i just did a initial commit of share-mem, a lib dedicated to deal with shared memory.
background : pd / pd~ communication is really slow. by example, having a pd~ patch with 8 audio in and 8 out, use about 50% cpu of a recent computer for each process only to deal with audio communication. sending a large array from one process to the other is almost not possible in RT. This really limit pd~ usability.
Thanks to share memory, communication between process can be greatly improve. This lib is mainly composed of an external, and few abstraction and examples.
Everything look stable, at least on a ubuntu (12.04) and a osX laptop. Implementation follow POSIX standard, so it will unfortunately not work on windows.
everything that I need is there, even if lot's more work could be made. I hope to have user feedback before making more development.
Cheers Cyrille
Pd-announce mailing list Pd-announce@iem.at http://lists.puredata.info/listinfo/pd-announce
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Hi,
Cyrille Henry wrote:
i just did a initial commit of share-mem, a lib dedicated to deal with shared memory.
Wow, great !
I'd like to use this to share pixes with veejay (a vjing tool that runs on linux). Veejay wants YUV and can read/write from/to shm. So it looks like communication can be done pretty easily. (note there's also an OGRE plugin that uses shm).
My question is: how can I map the shm id of pix_share_write to the output of ipcs ? Reading the source code left me clueless.
For example: [ set 400 256 256 RGBA ( creates this: ~$ ipcs key shmid size 0x0000798c 229379 charles 666 262164 1
veejay want the shm id as shown in the second field.
How course I can grep the pid and look for pd. But it won't work for several shms.
Thanks for any hints,
hello
looking at the thread, i was assuming that the question was about shmem, but it look more like a question about pix_share_write.
in shmem, the shmid is the id provide by the user so anyone can acces to the mem. but shmem did share table, so you have to put the pix in a table before sharing it.
on the other hand, pix_share_read / pix_share_write are dedicated to share pix data. but the id provide is not the shmid. looking at the source, i think the shmid is computed with this code :
int hash_str2us(std::string s) {
int result=0; int a=63689; int b=378551;
if(s.length()<1)return -1;
unsigned int i=0; for(i=0; i<s.length(); i++) { result=result*a+s[i]; a *= b; }
return ((unsigned short)(result) & 0x7FFFFFFF); }
is that what you where looking for?
cheers cyrille
Le 25/02/2013 15:13, Charles Goyard a écrit :
Hi,
Cyrille Henry wrote:
i just did a initial commit of share-mem, a lib dedicated to deal with shared memory.
Wow, great !
I'd like to use this to share pixes with veejay (a vjing tool that runs on linux). Veejay wants YUV and can read/write from/to shm. So it looks like communication can be done pretty easily. (note there's also an OGRE plugin that uses shm).
My question is: how can I map the shm id of pix_share_write to the output of ipcs ? Reading the source code left me clueless.
For example: [ set 400 256 256 RGBA ( creates this: ~$ ipcs key shmid size 0x0000798c 229379 charles 666 262164 1
veejay want the shm id as shown in the second field.
How course I can grep the pid and look for pd. But it won't work for several shms.
Thanks for any hints,
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 2013-02-25 15:54, Cyrille Henry wrote:
on the other hand, pix_share_read / pix_share_write are dedicated to share pix data. but the id provide is not the shmid.
right, it is a hashed version of the shmid, in order to make it more easily usable within Pd.
I'd like to use this to share pixes with veejay (a vjing tool that runs on linux). Veejay wants YUV and can read/write from/to shm. So it looks like communication can be done pretty easily. (note there's also an OGRE plugin that uses shm).
the thing with shm is, that you are sharing raw memory. just because multiple applications support reading/writing from/to shared memory, doesn't necessarily mean that they can share data in a meaningful way. the applications have to interpret the raw memory in the same way in order to make sense.
a standardized way for sharing videos between applications on linux is the "v4l2loopback" virtual video device.
fgmasdr IOhannes
Le 25/02/2013 16:17, IOhannes m zmoelnig a écrit :
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 2013-02-25 15:54, Cyrille Henry wrote:
on the other hand, pix_share_read / pix_share_write are dedicated to share pix data. but the id provide is not the shmid.
right, it is a hashed version of the shmid, in order to make it more easily usable within Pd.
i'm curious : what make the hash shmid more usable than the integer Id? i used the int id for shmem because i did not find any good reason not to.
cheers c
I'd like to use this to share pixes with veejay (a vjing tool that runs on linux). Veejay wants YUV and can read/write from/to shm. So it looks like communication can be done pretty easily. (note there's also an OGRE plugin that uses shm).
the thing with shm is, that you are sharing raw memory. just because multiple applications support reading/writing from/to shared memory, doesn't necessarily mean that they can share data in a meaningful way. the applications have to interpret the raw memory in the same way in order to make sense.
a standardized way for sharing videos between applications on linux is the "v4l2loopback" virtual video device.
fgmasdr IOhannes -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.12 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAlErgI4ACgkQkX2Xpv6ydvT93wCffiKg5+NmgRdV3+m2cJo9/N7e Z3wAoKGm8olsYLA/4Q/DhBj+N7SVEWAs =jd1H -----END PGP SIGNATURE-----
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 2013-02-25 16:43, Cyrille Henry wrote:
Le 25/02/2013 16:17, IOhannes m zmoelnig a écrit : On 2013-02-25 15:54, Cyrille Henry wrote:
on the other hand, pix_share_read / pix_share_write are dedicated to share pix data. but the id provide is not the shmid.
right, it is a hashed version of the shmid, in order to make it more easily usable within Pd.
i'm curious : what make the hash shmid more usable than the integer Id? i used the int id for shmem because i did not find any good reason not to.
without checking the actual code: the shmid can be any 32bit integer, e.g. 156823457 since Pd uses 32bit single precision floats, the given id might not be representable by a Pd-atom.
the hash i'm using is only 16bit, and thus always representable by a Pd-atom.
ftgmasdr IOhannes
ok, thanks for the explanation cheers c
Le 25/02/2013 16:48, IOhannes m zmoelnig a écrit :
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 2013-02-25 16:43, Cyrille Henry wrote:
Le 25/02/2013 16:17, IOhannes m zmoelnig a écrit : On 2013-02-25 15:54, Cyrille Henry wrote:
on the other hand, pix_share_read / pix_share_write are dedicated to share pix data. but the id provide is not the shmid.
right, it is a hashed version of the shmid, in order to make it more easily usable within Pd.
i'm curious : what make the hash shmid more usable than the integer Id? i used the int id for shmem because i did not find any good reason not to.
without checking the actual code: the shmid can be any 32bit integer, e.g. 156823457 since Pd uses 32bit single precision floats, the given id might not be representable by a Pd-atom.
the hash i'm using is only 16bit, and thus always representable by a Pd-atom.
ftgmasdr IOhannes -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.12 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAlErh9UACgkQkX2Xpv6ydvQD5ACdGKqjFP2lJolNwudvE98KbrVY ZZsAoO2z1R2MnSEhKg4txvkqUrpIKDsz =f6Zo -----END PGP SIGNATURE-----
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Hi,
Cyrille Henry wrote:
looking at the thread, i was assuming that the question was about shmem, but it look more like a question about pix_share_write.
Yes, it was, but since I failed to find the code of shmem (now I found it on svn), I assumed it was the same thing as pix_share_* (and then forgot about the general purpose of shmem). Sorry for the confusion.
Anyway, since I want to share video frames, pix_share sounds good.
Now as IOhannes points out, v4l2loopback is certainly more standard, and probably fast enough. So maybe I'll start with this and explore the other way round if I get a bottleneck. I can output to v4l with pix_record and the v4l2 backend, that's right ?
Thanks to all, Charles
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 2013-02-25 17:17, Charles Goyard wrote:
Hi,
Cyrille Henry wrote:
looking at the thread, i was assuming that the question was about shmem, but it look more like a question about pix_share_write.
Yes, it was, but since I failed to find the code of shmem (now I found it on svn), I assumed it was the same thing as pix_share_* (and then forgot about the general purpose of shmem). Sorry for the confusion.
Anyway, since I want to share video frames, pix_share sounds good.
Now as IOhannes points out, v4l2loopback is certainly more standard, and probably fast enough. So maybe I'll start with this and explore the other way round if I get a bottleneck. I can output to v4l with pix_record and the v4l2 backend, that's right ?
pix_record has backends for output to both v4l2 and v4l (aka: v4l1)
fgamndsr IOhannes
-
Antoine Villeret -
Charles Goyard -
Cyrille Henry -
Hans-Christoph Steiner -
IOhannes m zmoelnig -
Max -
Miller Puckette -
Nicolas Montgermont -
Phil Stone