Hans,
firstly, thanks a million for the hid device. it's something that is really appreciated.
...however, I am having some difficulties using it.
I just built a USB device which you can see here:
http://aug.ment.org/lechero
I was about to present it yesterday, but was having some major
problems with the [hid] object. I tested and built a patch with [hid] input on a desktop machine running gentoo, and it was working *perfectly*
- fast, stable and clean. I had it running all day and was editing a
running other patches at the same time. however, on my laptop running ubuntu, it was slow and cludgy, no matter how I set the polling. Whats also strange about this is that I changed the enumeration on my device so that 13 axis that I have show up properly in the registered abs_x, abs_y, etc. that you have in the 0.5 hid version. With the new enumeration, I can get [hid] to run and work perfectly on my desktop, however with the same code and with the same device, it crashes PD immediately when I turn the [hid] object on on my laptop. [a simple (if !=null) statement that I sent the other day will keep it from segfaulting, however. I just find it curious how it works on one machine but doesn't on another - I don't know much about USB]
so, for the presentation that was scheduled, I needed a working
solution on a laptop, ...and since everyone around me uses a mac, and since I thought you were developing [hid] for the mac, I decided to try it there. I downloaded your installer for PD with the hid object included, but to my surprise, the situation was even worse on the mac - the response from the [hid] object is _very_ slow and _very_ cludgy. I tried it on a coupla different macs, but got the same situ. In desperation I tried to set something up quick using supercollider on the mac, but the HIDdevice there was even worse.
I'd like to help debug if possible, but I don't really know where
to start.
thats my hid story so far.
-august.
A few thoughts offhand:
Is one linux box running a 2.4 kernel and the other 2.6? The [hid] 0.5
has some trouble with 2.6, I am working on that now.
Have you tried [linuxevent] or [linuxjoystick]? They are basically the
sketched for the [hid] object. They work pretty solidly, but have
limited feature set and work a fair amount differently that [hid].
I am pretty sure that the immediate segfaulting is caused by funky
element names. That I am fixing right now. Check out from CVS, it
might be there by the time you get up, but the anonymous CVS is about a
day behind.
I have never experienced the slowness thing that you are talking about.
Does it only affect the [hid] object/Pd? Try getting the data
directly from the device, i.e. cat /dev/input/event0. It'll look like
garbage, but you'll be able to see the activity.
.hc
On Jun 10, 2005, at 11:44 AM, august wrote:
Hans,
firstly, thanks a million for the hid device. it's something that is really appreciated.
...however, I am having some difficulties using it.
I just built a USB device which you can see here:
I was about to present it yesterday, but was having some major problems with the [hid] object. I tested and built a patch with [hid] input on a desktop machine running gentoo, and it was working
*perfectly*
- fast, stable and clean. I had it running all day and was editing a
running other patches at the same time. however, on my laptop running ubuntu, it was slow and cludgy, no matter how I set the polling. Whats also strange about this is that I changed the enumeration on my device
so that 13 axis that I have show up properly in the registered abs_x,
abs_y, etc. that you have in the 0.5 hid version. With the new enumeration, I can get [hid] to run and work perfectly on my desktop, however with the same code and with the same device, it crashes PD immediately when I
turn the [hid] object on on my laptop. [a simple (if !=null) statement
that I sent the other day will keep it from segfaulting, however. I just
find it curious how it works on one machine but doesn't on another - I don't
know much about USB]so, for the presentation that was scheduled, I needed a working solution on a laptop, ...and since everyone around me uses a mac, and since I thought you were developing [hid] for the mac, I decided to
try it there. I downloaded your installer for PD with the hid object
included, but to my surprise, the situation was even worse on the mac - the
response from the [hid] object is _very_ slow and _very_ cludgy. I tried it on
a coupla different macs, but got the same situ. In desperation I tried
to set something up quick using supercollider on the mac, but the
HIDdevice there was even worse.I'd like to help debug if possible, but I don't really know where to start.
thats my hid story so far.
-august.
PD-list@iem.at mailing list UNSUBSCRIBE and account-management ->
http://lists.puredata.info/listinfo/pd-list
"Computer science is no more related to the computer than astronomy is
related to the telescope."
-Edsger Dykstra
First off, I have to say, El Lechero looks very nice. I am eagerly
awaiting seeing it in action.
Upon re-reading this email, I am not entirely sure what I changed in
0.6 to fix it. ( I have some guesses). Is the slow, cludginess
totally gone?
Is your Desktop running a 2.4 kernel and your laptop a 2.6 kernel?
.hc
On Jun 10, 2005, at 11:44 AM, august wrote:
Hans,
firstly, thanks a million for the hid device. it's something that is really appreciated.
...however, I am having some difficulties using it.
I just built a USB device which you can see here:
I was about to present it yesterday, but was having some major problems with the [hid] object. I tested and built a patch with [hid] input on a desktop machine running gentoo, and it was working
*perfectly*
- fast, stable and clean. I had it running all day and was editing a
running other patches at the same time. however, on my laptop running ubuntu, it was slow and cludgy, no matter how I set the polling. Whats also strange about this is that I changed the enumeration on my device
so that 13 axis that I have show up properly in the registered abs_x,
abs_y, etc. that you have in the 0.5 hid version. With the new enumeration, I can get [hid] to run and work perfectly on my desktop, however with the same code and with the same device, it crashes PD immediately when I
turn the [hid] object on on my laptop. [a simple (if !=null) statement
that I sent the other day will keep it from segfaulting, however. I just
find it curious how it works on one machine but doesn't on another - I don't
know much about USB]so, for the presentation that was scheduled, I needed a working solution on a laptop, ...and since everyone around me uses a mac, and since I thought you were developing [hid] for the mac, I decided to
try it there. I downloaded your installer for PD with the hid object
included, but to my surprise, the situation was even worse on the mac - the
response from the [hid] object is _very_ slow and _very_ cludgy. I tried it on
a coupla different macs, but got the same situ. In desperation I tried
to set something up quick using supercollider on the mac, but the
HIDdevice there was even worse.I'd like to help debug if possible, but I don't really know where to start.
thats my hid story so far.
-august.
PD-list@iem.at mailing list UNSUBSCRIBE and account-management ->
http://lists.puredata.info/listinfo/pd-list
As we enjoy great advantages from inventions of others, we should be
glad of an opportunity to serve others by any invention of ours; and
this we should do freely and generously.
- Benjamin Franklin
First off, I have to say, El Lechero looks very nice. I am eagerly awaiting seeing it in action.
here is a short vid of a presentation I did:
http://www.mat.ucsb.edu/594O/finals05/August.mp4
Upon re-reading this email, I am not entirely sure what I changed in 0.6 to fix it. ( I have some guesses). Is the slow, cludginess totally gone?
as far as I can see, yes. however I haven't tried on a mac yet. Before, the same pd code running on a Mac G5 was almost 3 times as slow as on my linux box. I'll let you know how I make on a mac as soon as I get a chance to hook up to one.
Is your Desktop running a 2.4 kernel and your laptop a 2.6 kernel?
nope. all are 2.6 kernels.
-gust.
-
august -
Hans-Christoph Steiner