Hey, This is prolly the noobest question ever but i'm new and dunno how to get around linux that well.
How do i get flow_objects.c from gridflow/base/ running like seen in this link: http://gridflow.ca/gallery/multiblob.png
Thnks, Sisil
On Tue, 4 Aug 2009, sisil mehta wrote:
This is prolly the noobest question ever but i'm new and dunno how to get around linux that well. How do i get flow_objects.c from gridflow/base/ running like seen in this link:
flow_objects.c is just a source file containing most builtin object classes of gridflow. you don't use it directly, you run "make" to make gridflow.pd_linux from that file (and from other files all together). Once you have successfully loaded the "gridflow" library in Pd, all builtins are supposed to be available. You shouldn't have any "undefined symbol" error, because if you do, you're sure gridflow.pd_linux is NOT loaded.
almost every gridflow patch is using "flow_objects.c". maybe I shouldn't have mentioned that file in the first place.
please make sure that you have gridflow loading properly at first, perhaps by loading examples/hello-world.pd first, to see whether anything works at all. There's no reason why [#labeling] wouldn't work.
Check the spelling, because I spelt [#labeling] with a single l, instead of the double l such as "labelling" (both exist in english).
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
Hey, ya, gridflow does work, i tried some examples . so to get the result as shown in : http://gridflow.ca/gallery/multiblob.png(the multiblob patch ) do i have to write the program for it? so is it available somewhere? Sisil.
On Wed, Aug 5, 2009 at 6:07 AM, Mathieu Bouchard matju@artengine.ca wrote:
On Tue, 4 Aug 2009, sisil mehta wrote:
This is prolly the noobest question ever but i'm new and dunno how to get
around linux that well. How do i get flow_objects.c from gridflow/base/ running like seen in this link:
flow_objects.c is just a source file containing most builtin object classes of gridflow. you don't use it directly, you run "make" to make gridflow.pd_linux from that file (and from other files all together). Once you have successfully loaded the "gridflow" library in Pd, all builtins are supposed to be available. You shouldn't have any "undefined symbol" error, because if you do, you're sure gridflow.pd_linux is NOT loaded.
almost every gridflow patch is using "flow_objects.c". maybe I shouldn't have mentioned that file in the first place.
please make sure that you have gridflow loading properly at first, perhaps by loading examples/hello-world.pd first, to see whether anything works at all. There's no reason why [#labeling] wouldn't work.
Check the spelling, because I spelt [#labeling] with a single l, instead of the double l such as "labelling" (both exist in english).
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
On Wed, 5 Aug 2009, sisil mehta wrote:
http://gridflow.ca/gallery/multiblob.png(the multiblob patch ) do i have to write the program for it? so is it available somewhere? Sisil.
I'm finally adding examples/multiblob.pd based on the screenshot of occupants-video.pd found in that screenshot. However I cut out the mask made by [# &] for columns 312-319, because this was specifically for a camera that had a lot of noise in those columns.
I'm also merging in some parts that were never seen, from later versions of the patch (after the screenshot that you know). One of the "new" features is the rotation of the crosses.
I'm also cutting off some useless things from that patch, and I don't just mean the parts marked as "unused", I mean object combinations that were detours that existed only because of the way I made the patch in the first place.
Eventually, I might replace the thresholder by a complete colour detector.
new screenshot at http://gridflow.ca/gallery/multiblob2.png
new patch is at http://gridflow.ca/svn/trunk/examples/multiblob.pd (just use svn update)
I made some changes after making the screenshot: crosses aren't square anymore (change in #make_cross.pd) and the selected areas are now dotted white with some kind of 1/6 stipple.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
Hi @mathieu, Thanks a lot man...u just saved my internship....owe you one. Sisil.
On Mon, Aug 17, 2009 at 1:37 AM, Mathieu Bouchard matju@artengine.cawrote:
On Wed, 5 Aug 2009, sisil mehta wrote:
http://gridflow.ca/gallery/multiblob.png(thehttp://gridflow.ca/gallery/multiblob.png%28themultiblob patch ) do i have to write the program for it? so is it available
somewhere? Sisil.
I'm finally adding examples/multiblob.pd based on the screenshot of occupants-video.pd found in that screenshot. However I cut out the mask made by [# &] for columns 312-319, because this was specifically for a camera that had a lot of noise in those columns.
I'm also merging in some parts that were never seen, from later versions of the patch (after the screenshot that you know). One of the "new" features is the rotation of the crosses.
I'm also cutting off some useless things from that patch, and I don't just mean the parts marked as "unused", I mean object combinations that were detours that existed only because of the way I made the patch in the first place.
Eventually, I might replace the thresholder by a complete colour detector.
new screenshot at http://gridflow.ca/gallery/multiblob2.png
new patch is at http://gridflow.ca/svn/trunk/examples/multiblob.pd (just use svn update)
I made some changes after making the screenshot: crosses aren't square anymore (change in #make_cross.pd) and the selected areas are now dotted white with some kind of 1/6 stipple.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
On Mon, 17 Aug 2009, sisil mehta wrote:
@mathieu, Thanks a lot man...u just saved my internship....owe you one.
BTW, take a white wall, shine a bright light all over it diagonally, and not on anything else. walk in front of the wall. you will appear relatively quite dark to the camera.
then use [# <] instead of [# >] to select dark spots instead of bright ones.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
hey, In the multiblob.pd pactch that u addded (i got it ffrom svn) i cant even open it. PD crashes when i try to open the patch(infact i cant even see the window with the patch when i click on multiblob.pd). Is the camera enabled by default in this patch? (because i got a similar error when using the #camera module, crashing PD). Could u not include the #camera section? I wa thinking of getting a .mov file into te patch and trying outt he multiblob algo on it instead of a video feed from the camera(i can add the .mov part ). or should i just use the algo as seen the screen shot u gave?
Sisil.
On Mon, Aug 17, 2009 at 1:37 AM, Mathieu Bouchard matju@artengine.cawrote:
On Wed, 5 Aug 2009, sisil mehta wrote:
http://gridflow.ca/gallery/multiblob.png(thehttp://gridflow.ca/gallery/multiblob.png%28themultiblob patch ) do i have to write the program for it? so is it available
somewhere? Sisil.
I'm finally adding examples/multiblob.pd based on the screenshot of occupants-video.pd found in that screenshot. However I cut out the mask made by [# &] for columns 312-319, because this was specifically for a camera that had a lot of noise in those columns.
I'm also merging in some parts that were never seen, from later versions of the patch (after the screenshot that you know). One of the "new" features is the rotation of the crosses.
I'm also cutting off some useless things from that patch, and I don't just mean the parts marked as "unused", I mean object combinations that were detours that existed only because of the way I made the patch in the first place.
Eventually, I might replace the thresholder by a complete colour detector.
new screenshot at http://gridflow.ca/gallery/multiblob2.png
new patch is at http://gridflow.ca/svn/trunk/examples/multiblob.pd (just use svn update)
I made some changes after making the screenshot: crosses aren't square anymore (change in #make_cross.pd) and the selected areas are now dotted white with some kind of 1/6 stipple.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
sisil mehta wrote:
hey, In the multiblob.pd pactch that u addded (i got it ffrom svn) i cant even open it. PD crashes when i try to open the patch(infact i cant even see the window with the patch when i click on multiblob.pd). Is the camera enabled by default in this patch? (because i got a similar error when using the #camera module, crashing PD). Could u not include the #camera section? I wa thinking of getting a .mov file into te patch and trying outt he multiblob algo on it instead of a video feed from the camera(i can add the .mov part ). or should i just use the algo as seen the screen shot u gave?
you could open the patch without GF being loaded: it won't work but it shouldn't crash then and you can remove the objects you think are causing the crash all by yourself.
fgamsdr IOhannes
Hey,
i edited multiblob.pd and removed the camera source. Instead i added violoncella.mov as a source (play_video.pd which runs violoncella.mov works fine.) Now i ran the new multiblob.pd (i've attahced the file) and it doesnt run. PD crashes giving the following errors: in dmesg: CE: hpet increasing min_delta_ns to 22500 nsec pd[9542]: segfault at 4 ip b76be5dc sp bf9eb4f0 error 6 in libstdc++.so.6.0.10[b7605000+e4000] pd[9693]: segfault at 4 ip b77e45dc sp bff11a10 error 6 in libstdc++.so.6.0.10[b772b000+e4000]
using gdb: Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0xb7da26c0 (LWP 10295)] 0xb77035dc in __cxa_allocate_exception () from /usr/lib/libstdc++.so.6
using the where command , i've attahced the result.
I've been trying to run something on this for the last 2 weeks. I still havent been able to. I had issues with Gem also so the whole Gem2GF didnt work out. Even now when i run a Gem application, it runs but when i try to destroy the window pd crashes. Hasnt anyone experienced this before??? Is it a issue with the graphics card(ATI Radeon 4570)?? should i just change the linux distro(Ubuntu 9.04)?? Has anyone used a distro where all these features have worked??
Sisil.
On Mon, Aug 17, 2009 at 9:35 PM, IOhannes m zmoelnig zmoelnig@iem.atwrote:
sisil mehta wrote:
hey, In the multiblob.pd pactch that u addded (i got it ffrom svn) i cant even open it. PD crashes when i try to open the patch(infact i cant even see the window with the patch when i click on multiblob.pd). Is the camera enabled by default in this patch? (because i got a similar error when using the #camera module, crashing PD). Could u not include the #camera section? I wa thinking of getting a .mov file into te patch and trying outt he multiblob algo on it instead of a video feed from the camera(i can add the .mov part ). or should i just use the algo as seen the screen shot u gave?
you could open the patch without GF being loaded: it won't work but it shouldn't crash then and you can remove the objects you think are causing the crash all by yourself.
fgamsdr IOhannes
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
On Mon, 17 Aug 2009, sisil mehta wrote:
i edited multiblob.pd and removed the camera source. Instead i added violoncella.mov as a source (play_video.pd which runs violoncella.mov works fine.) Now i ran the new multiblob.pd (i've attahced the file) and it doesnt run. PD crashes giving the following errors:
I hope that you have spelt that name right in the actual patch, because in this email, you didn't. It's "violoncelle.mov". this is french spelling for "cello".
0xb77035dc in __cxa_allocate_exception () from /usr/lib/libstdc++.so.6
Nearly every error message of GridFlow is turned into a crash in __cxa_allocate_exception because of the ATI/C++ driver conflict. There might be a GDB command that can show the contents of the error message in such a case, but i don't recall how...
I can't test the ATI/C++ driver conflict because I don't have access to such a computer these days.
I've been trying to run something on this for the last 2 weeks. I still havent been able to. I had issues with Gem also so the whole Gem2GF didnt work out. Even now when i run a Gem application, it runs but when i try to destroy the window pd crashes. Hasnt anyone experienced this before??? Is it a issue with the graphics card(ATI Radeon 4570)?? should i just change the linux distro(Ubuntu 9.04)?? Has anyone used a distro where all these features have worked??
AFAIK, the ATI driver problem occurs regardless of the distro. Perhaps someone can tell whether there is an alternate ATI driver from a different author, or any history of which versions have this conflict...
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
On Mon, 17 Aug 2009, Mathieu Bouchard wrote:
I hope that you have spelt that name right in the actual patch, because in this email, you didn't. It's "violoncelle.mov". this is french spelling for "cello".
Sorry, I didn't look at your log when I first saw that. The log shows that you had spelt it right. It also corresponds to this error message I get when running my own patch:
error: [#draw_polygon put (255 0 0)]: 'lines' is empty
which is an initialisation error I made, but didn't worry about because it doesn't cause any problem (except with the ATI driver).
0xb77035dc in __cxa_allocate_exception () from /usr/lib/libstdc++.so.6
Can anyone figure out how to fix this problem from within GridFlow? Because so far, the solutions involve messing with pd config in weird ways and I can't expect every GF user will be experienced and/or persistent enough to figure it out. What I know about it is essentially what Claude wrote in this mail:
http://lists.artengine.ca/pipermail/gridflow-dev/2008-October/000455.html
And if you read in the gridflow-dev archive that this bug has to do with Ubuntu, it was a mistake, it doesn't (and I replied to myself about it too).
I thought about something: could anyone with that problem try to recompile Pd with a -lstdc++ option? if it worked that way, there would be no need for the library loading order hack, and no need for the even weirder hack involving a dummy "external" that contains nothing (the latter hack is required for those who want the GEM<->GF bridge, because of some kind of catch-22).
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
Hey,
On Tue, Aug 18, 2009 at 1:45 AM, Mathieu Bouchard matju@artengine.cawrote:
On Mon, 17 Aug 2009, Mathieu Bouchard wrote:
I hope that you have spelt that name right in the actual patch, because in this email, you didn't. It's "violoncelle.mov". this is french spelling for "cello".
Sorry, I didn't look at your log when I first saw that. The log shows that you had spelt it right. It also corresponds to this error message I get when running my own patch:
error: [#draw_polygon put (255 0 0)]: 'lines' is empty
which is an initialisation error I made, but didn't worry about because it doesn't cause any problem (except with the ATI driver).
0xb77035dc in __cxa_allocate_exception () from /usr/lib/libstdc++.so.6
Can anyone figure out how to fix this problem from within GridFlow? Because so far, the solutions involve messing with pd config in weird ways and I can't expect every GF user will be experienced and/or persistent enough to figure it out. What I know about it is essentially what Claude wrote in this mail:
I checked out this link:
http://lists.artengine.ca/pipermail/gridflow-dev/2008-October/000455.html
What do i have to do with the attachments?? in the terminal i just executed this: (i dwnld attachment.c and renamed it libstdcpp.c) gcc -shared -o libstdcpp.pd_linux libstdcpp.c -lstdc++ and got this: /usr/bin/ld: cannot find -lstdc++ collect2: ld returned 1 exit status
And if you read in the gridflow-dev archive that this bug has to do with Ubuntu, it was a mistake, it doesn't (and I replied to myself about it too).
I thought about something: could anyone with that problem try to recompile Pd with a -lstdc++ option? if it worked that way, there would be no need for the library loading order hack, and no need for the even weirder hack involving a dummy "external" that contains nothing (the latter hack is required for those who want the GEM<->GF bridge, because of some kind of catch-22).
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
On Tue, 18 Aug 2009, sisil mehta wrote:
I checked out this link:
http://lists.artengine.ca/pipermail/gridflow-dev/2008-October/000455.html
What do i have to do with the attachments?? in the terminal i just executed this: (i dwnld attachment.c and renamed it libstdcpp.c)
seems right.
gcc -shared -o libstdcpp.pd_linux libstdcpp.c -lstdc++ and got this: /usr/bin/ld: cannot find -lstdc++ collect2: ld returned 1 exit status
If -lstdc++ didn't work then how could you compile GridFlow at all?
try this: ls -l /usr/lib/libstdc++.* I get this (I cut off the left part of the listing):
18 2009-05-08 19:42 /usr/lib/libstdc++.so.5 -> libstdc++.so.5.0.7
737192 2008-05-10 02:18 /usr/lib/libstdc++.so.5.0.7 19 2009-05-09 21:09 /usr/lib/libstdc++.so.6 -> libstdc++.so.6.0.10 950424 2009-03-16 20:28 /usr/lib/libstdc++.so.6.0.10
In the meanwhile, I made a video of violoncelle.mov in multiblob.pd :
http://gridflow.ca/gallery/violoncelle_multiblob.mov
For this I used the [# >] threshold with some value I picked to make it visually interesting. (but practical uses would replace [# >] by something else.)
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
On Tue, Aug 18, 2009 at 2:47 AM, Mathieu Bouchard matju@artengine.cawrote:
On Tue, 18 Aug 2009, sisil mehta wrote:
I checked out this link:
http://lists.artengine.ca/pipermail/gridflow-dev/2008-October/000455.html
What do i have to do with the attachments?? in the terminal i just executed this: (i dwnld attachment.c and renamed it libstdcpp.c)
seems right.
gcc -shared -o libstdcpp.pd_linux libstdcpp.c -lstdc++ and got this:
/usr/bin/ld: cannot find -lstdc++ collect2: ld returned 1 exit status
If -lstdc++ didn't work then how could you compile GridFlow at all?
try this: ls -l /usr/lib/libstdc++.* I get this (I cut off the left part of the listing):
i tried the above stment and got htis: lrwxrwxrwx 1 root root 19 2009-08-06 22:37 /usr/lib/libstdc++.so.6 -> libstdc++.so.6.0.10 -rw-r--r-- 1 root root 950424 2009-03-17 05:58 /usr/lib/libstdc++.so.6.0.10
After that i tried some examples in Gem (the libstdcpp lib loaded sucessfully.) and it crashed with a new error:
segfault at b65c8a52 ip b65c8a52 sp bfc7dbe0 error 4 in libgomp.so.1.0.0[b6a19000+7000] segfault at b7bc3014 ip b7f27b64 sp bf995624 error 7 in libpthread-2.9.so [b7f21000+15000]
i tried to recompile gridflow ...and noticed the following error:
[objcpp] GNU/Apple ObjectiveC++ Compiler: ------------------------------------> missing (where is objc/Object.h ?) (*I have dwnlded and installed gnu objective c++ compiler: gobjc++ but it still gives this*) [gem09] PureData GEM (source code) with m_holdname: --------------------------> missing (gcc: error: ‘CPPExtern’ has not been declared) [gem08] PureData GEM (source code) without m_holdname: -----------------------> missing (gcc compilation error)
could this be an issue?
Sisil.
18 2009-05-08 19:42 /usr/lib/libstdc++.so.5 -> libstdc++.so.5.0.7737192 2008-05-10 02:18 /usr/lib/libstdc++.so.5.0.7 19 2009-05-09 21:09 /usr/lib/libstdc++.so.6 -> libstdc++.so.6.0.10 950424 2009-03-16 20:28 /usr/lib/libstdc++.so.6.0.10
In the meanwhile, I made a video of violoncelle.mov in multiblob.pd :
http://gridflow.ca/gallery/violoncelle_multiblob.mov
For this I used the [# >] threshold with some value I picked to make it visually interesting. (but practical uses would replace [# >] by something else.)
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
On Mon, 17 Aug 2009, IOhannes m zmoelnig wrote:
you could open the patch without GF being loaded: it won't work but it shouldn't crash then and you can remove the objects you think are causing the crash all by yourself.
It's an alternative to -noloadbang but I'd say -noloadbang is a bit easier as you don't have to deconfigure anything.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
Mathieu Bouchard wrote:
On Mon, 17 Aug 2009, IOhannes m zmoelnig wrote:
you could open the patch without GF being loaded: it won't work but it shouldn't crash then and you can remove the objects you think are causing the crash all by yourself.
It's an alternative to -noloadbang but I'd say -noloadbang is a bit easier as you don't have to deconfigure anything.
definitely. since i didn't know wha tactually caused the problem (whether it was a [loadbang] or an object initialization by arguments), i suggested the safe (but more complex way).
so in short: if a patch crashes, try:
- disabling loadbang by using the "-noloadbang"
OR (if the above fails)
- make Pd not find crashy objects, by not loading libraries, not giving
paths,...
fgmasdr IOhannes
On Mon, 17 Aug 2009, sisil mehta wrote:
In the multiblob.pd pactch that u addded (i got it ffrom svn) i cant even open it. PD crashes when i try to open the patch(infact i cant even see the window with the patch when i click on multiblob.pd). Is the camera enabled by default in this patch? (because i got a similar error when using the #camera module, crashing PD).
Yes, [#camera 0] opens device #0, "/dev/video0". The default value of $1 of #camera is -1, which means "OFF" (nothing open).
Could u not include the #camera section?
Usually, you'd use the startup option -noloadbang to get the patch loaded. [#camera] and [#in] are abstractions which both use plain [loadbang]s to do their stuff. This should allow you to edit the patch, save, restart pd without -noloadbang and then run it.
I wa thinking of getting a .mov file into te patch and trying outt he multiblob algo on it instead of a video feed from the camera(i can add the .mov part ). or should i just use the algo as seen the screen shot u gave?
the new screenshot has more hidden parts (subpatches) and also it's not the latest version of the patch, as I said.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec
-
IOhannes m zmoelnig -
Mathieu Bouchard -
sisil mehta