I think you misread my technical statement as a political one. I
don't care about politics or the GPL, I just want Linux to be the most
stable OS, and that can't happen if secret blobs of code are allowed
to scribble all over kernel memory.
I have an additional argument against binary
spam. My idea is to find a way to tweak (or apply an existing patch
to) mailman so that messages marked as spam get discarded
automatically
mailman on my mailserver (Ubuntu Dapper out of the box) has this option:
Privacy options...
[Spam filters]
Header filters
Filter rules to
not sure what you mean, but on sparcs, int writes are not atomic unless
you only use the lower 24 bits.
right, this is exactly the point! 32 bits int might be atomic on a certain
platform, but i have no idea about many others.
in the meantime, a part from joq suggestion of LFDS, and niklas
Hello,
I am looking for a cross-platform implementation of an atomic
integer.
Under Linux, a build an c++ class atomic around asm/atomic.h,
(which I can use as if it where an int), but I'd like to have a
solution that also works on Windows XP and Mac OS X.
Thanks for any suggestions,
Maarten
Very interesting reads. Thanks!
Here's some papers specifically geared toward DSP processors that
support the use of fixed point:
Superior Audio Requires Fixed-Point DSP
http://www.rane.com/note153.html
48-Bit Integer Processing Beats 32-Bit Floating-Point for Professional
Audio
Hi,
Thanks for the reply.
CONFIG_PREEMPT_RT: Complete Preemption (Real-Time)
[...]
AFAIK this one.
Hm, I am currently trying PREEMPT_DESKTOP.
I got the impression _RT would be a bit of overkill for audio work.
Unfortunately, I also try to get decent audio on my 2.6.12
machine but
List archives. There's no point documenting it because it's developing
too fast.
Ok. I gave up following the whole discussing on LKML because it was too
fast :-)
If your distro has a 2.6.12 kernel available, you need to file a bug
report against PAM to support this new kernel feature.
I
hello,
any advice on what would be the best kernel options
when using ingo molnar's patch for an audio setup?
CONFIG_PREEMPT_DESKTOP: Preemptible Kernel (Low-Latency Desktop)
or
CONFIG_PREEMPT_RT: Complete Preemption (Real-Time)
CONFIG_PREEMPT_SOFTIRQS: Thread Softirqs
Hello,
I just did some rereading of some parts of the What Parts of Linux
Audio Simply Work Great? thread, that talk about the problems with
soundcards that do not support multiple streams, and thought it would be
good if we could actually come up with an advice to the desktop
developers (Gnome
Hi Lee,
why alsa does not
do mixing in software by default
Please reread the thread again. It does do this by default. ALSA 1.0.9
or later is required.
I knew this. But in the context of explaining the problem, it's essential
to mention it. I should have said did instead of does.
All
Paul Davis [EMAIL PROTECTED] wrote:
i'll leave making things as fast as possible to intel, AMD and the gcc team.
When I read this when you posted this the first time, I thought: and
what about PPC? But now, after Steve Jobs WWDC keynote, it turns out
that your omission of PPC was visionary :-)
interest in the AES crowd, and given them a good idea of the current
state of Linux audio applications.
We put the slides at http://iua-share.upf.es/wikis/aes/ and tomorrow we
will add some photos. Of course suggestions are welcome!
Pau Arumi
Maarten de Boer
Hello,
I decided to give kernel 2.6.10 a try, so I installed it in combination
with Con Kolivas system system responsiveness patch. But when I try to
run the latencytest (0.5.5), I run into the following problem:
$ modprobe latency-test
WARNING: Error inserting rtc
Hi Takashi,
Thanks for your reply.
Did you enable HPET?
Hmm, it seems so... I configured my kernel using the debian .config
as oldconfig.
$ grep HPET /boot/config-2.6.10-ck2-pentium4
CONFIG_HPET_TIMER=y
CONFIG_HPET=y
# CONFIG_HPET_RTC_IRQ is not set
CONFIG_HPET_MMAP=y
When hpet timer is
Hello,
I decided to give the voluntary preempt patch a try, so I just build a
2.6.9-rc3-mm3-VP-T3 kernel. But when I try to run the latency-test I
run into the following problems:
First of all, a simple-to-fix compile error in showtrace.c, caused by a
missing
while ((c = getopt(argc,
Hello Florian,
Is this with kernel 2.6.x or 2.4.x?
I remember there were some problems with Matrox cards, that were
initially not included in the 2.4.x low-latency patch, but I thought
that at some point Andrew included them.
http://www.eca.cx/lad/2002/06/0076.html
maarten
[...] I didn't have the original MS TTF fonts anywhere on my system,
so I grabbed them from my Better Half's machine
For those of you that don't have a Better Half (that runs Windows), you
can download these fonts from the http://corefonts.sourceforge.net/
download pages.
You also need
Hello,
[EMAIL PROTECTED] wrote:
And did you find RoseGarden and Ardour we're sync'd up tightly?
I have yet to find them REALLY syncing really tightly
what distro are you on? what versions of the software? what hardware?
Sorry for not replying earlier. We wanted to do some tests
This combination won't work for synchronised MIDI and audio with
Rosegarden.
Really? It seems to have worked for us. What are the symptoms?
Hello,
I am preparing some slides about Linux audio, and while comparing Linux
with Windows, I have been wondering how the ASIO drivers manage to
obtain low latency on MS Windows, an operating system that does not seem
capable of low latency in any other way. So what tricks did Steinberg
come up
Hi Erik,
Depending on the ranges of your increment, and the accuracy you
want to obtain, you might consider doing this with integers only.
Maarten
The fix in this case was this:
for (;;)
{
/* Bunch of other code. */
fractional += increment ;
rem = fmod
Hi,
This subject reminded me that some time ago, a friend of mine mentioned
that he had read a message of James McCartney (SuperCollider) in which
James talked about his decision of having implemented his own OO
language for SC rather than reusing an existing, generic, OO language. I
looked for
a collegue pointed me to this post about a high resolution time
patch for the 2.6.5 kernel. i'm not 100% if it has been mentioned
here before, nor of how much interest it is, but just in case.
http://lwn.net/Articles/81400/
maarten
debian is about to exclude all non-free material from their
distributions. this includes firmware, documentation et al.
Hi Paul,
Where did you read that? It is not what i understand from the changes to
the Social Contract, http://www.debian.org/vote/2004/vote_003
The problem is that the guy that took the video used a camcorder with
firewire output, he is not a Linux expert, he uses windows on his PC
which has very to easy tools to transfer the videos to the PC and make a
video file out of it.
On Linux, kino will do this just fine, with a nice GUI.
Begin forwarded message:
The Music Technology Group of the Universitat Pompeu Fabra in Barcelona
(www.iua.upf.es/mtg), a research group known for its work on audio
processing technologies and their musical and multimedia applications,
has 4 open positions in different areas and projects:
-
hello,
i remember that some time ago there was some discussion about
bandwidth limited waveform generators. my question is: did
anybody compare the various implementations, and does anybody
have an idea what would be the fastest? (square, saw)
maarten
Thanks!
Out of curiosity: how did you manage to keep binary compatibility?
Maarten
oh it'll be audible, and as a lead vocal, you can bet it'll sound cheesy as
hell, except for that FIRST person who uses it, which will probably be
Prince
but it might be useful for adding backup singers - which is actually what
weirds me out -
and during the composition phase, just to get
[snip: paul davis' gtk embedding example]
well, looking at your code, i am not really sure if this is what you
mean, but the attached code shows how to embed an existing X11 window
inside a fltk window (run it, find out the XId of any existing window
(with xwininfo), and enter it in the text
For what it's worth, this can also be done with FLTK.
Maarten
XEMBED style, or foreign window style ? i've already learnt how to
do it with Qt ...
I don't remember.. I did it once. I think it was foreign window style.
Anyway, I'll give it a try tomorrow.
Maarten
you might also want to check out Gray Scavone's RtAudio, which is C++
where PortAudio is C, and it provides both blocked and callback based
I/O.
RtAudio is a C++ class which provides a common API (Application
Programming Interface) for realtime audio input/output across Linux
(native ALSA and
By the way, it seems the PPC has instructions to do it.
From the PPC compiler writers guide:
Round Up or Down to Next Power of 2
The floor power of 2 ( flp2) and ceiling power of 2 ( clp2) functions
are similar to the floor and ceiling functions, respectively, but they
round to an integral power
PS: I couldn't tell from the list guidelines whether this kind of post
is appropriate. If not, sorry! and please ignore it.
I think that the general consensus is that this kind of post are welcome.
Linux audio jobs, or even computer audio jobs, are rare enough, and an
employer has little other
Hi,
A colleague of mine found this very clever algoritm to calculate the
next power of two for a given number. It comes from the Prophecy SDK for
3d game development
http://www.twilight3d.com/modules.php?op=modloadname=Downloadsfile=index
Since this is a kind of thing often needed in audio
This is a nice coincidence, because I just installed Debian Sid
('unstable') on my home workstation.
I am also a convinced Debian user, and all the servers and workstations
I administrate are Debian Woody ('stable'). Now, recently, I have seen
myself forced to install more and more backported
By the way, Knoppix uses Debian sid, and is overall considered to be a
very good demonstration of Linux' capabilities, both in features and in
stability. I don't think it would have gotten the recognition it has, if
people would have found it buggy or unstable.
Maarten
for the record, plain old nfs has always worked for me. any reasons you
are not just using that ? (ok, i know it sucks in many respects,
esepcially security, but then low latency audio implied a trusted
environment anyway...)
yes, nfs is what we are using currently. the problem is that
We have the same situation at work, I have root on my desktop so I'm not
allowed to NFS mount the fileserver.
But it will allow you to su to any user id, and mount it. That's the
problem. AUTH_DES seems to be the solution, but is not implemented on
Linux :-(
Have you looked at shfs, it allows
http://shfs.sourceforge.net/
LUFS looks quite a bit more mature.
http://lufs.sourceforge.net/lufs/
Hello,
I have been doing some tests with OpenAFS, and it hangs on me
frequently. On the OpenAFS I found out that this is very likely
caused by the preemptible patch. Knowing that a lot of people
around here are using that patch, I'd like to ask if any of you
have experienced the same, and know of
I experienced the same problem. I posted on OpenAFS-info but nobody
came up with a solution or even indicated that the problem had been
looked into at all.
Well, I got a reaction that was even less encouraging:
Well, it's either AFS or the preemption patch. You can choose...
Maarten
i'd encourage people to try 2.5 to iron out audio-specific problems
before we go into 2.6-test and the developers get nervous...
I'd love to, but a real show-stopper is that the (evil closed source)
nvidia drivers don't support the 2.5 kernel... I've seen some patches
for older versions
Has anybody got the (latest) NVidia drivers to
work with the latest 2.5 kernel?
Ah, answering myself, I just found patches at http://www.minion.de/
Maarten
look here:
http://members.optusnet.com.au/ckolivas/kernel/
thanks, that's a very interesting page.
maarten
hello,
it seems that the preemptible and the low latency patches are not yet
available for the latest kernel, 2.4.21
has anybody succesfully applied older patches? any idea if the new
patches are being worked on?
maarten
and that seems strange? ;-)
no, though in the past i have been amazed sometimes :-)
the -ac changelog. To me, it's trivial to patch the rejects, but to
someone that isn't all that excited about it, it can be daunting.
sure, but getting the rejects to patch does not mean that the
patch is
On Thu, 6 Mar 2003 18:46:27 +0100
Maarten de Boer [EMAIL PROTECTED] wrote:
i have not tried this, but i would say that using mencoder in copy mode
writing to a fifo and mplayer reading from that fifo could do the trick.
I was talking nonsense here.
Maarten
i have not tried this, but i would say that using mencoder in copy mode
writing to a fifo and mplayer reading from that fifo could do the trick.
the time between launching mencoder and mplayer determines that delay.
maarten
Hello,
I just released polarbear. I had the code lying around, and just merged
it with the jack/alsa i/o code of tapiir. Note that this is the first
public release. I did not test it thoroughly, and I am not sure if the
GUI is obvious enough (it should be if you are familiar with complex
Hello,
I just released a new version of tapiir. Tapiir now supports jack.
Tapiir can be found at
http://www.iua/~mdeboer/projects/tapiir/
Tapiir is a simple and flexible audio effects processor, inspired on
the classical magnetic tape delay systems used since the early days
of electro-acoustic
I couldn't resist it so I hacked up a quick script to try the blockless,
dynamicly compiled processing we were discussing the other day.
Hi,
I missed that discussion (do you have a pointer?) but it all seems very
much related to the example i posted some while ago:
my thought was that i could avoid this by actually emitting SMPTE
itself, which is an audio stream, and thus can be generated with
perfect sample accuracy as we go. this can then allow people with
equipment that can read a SMPTE signal and/or convert it to MTC to get
their devices to lock to
i know that someone posted smpte-reading code here last year. does
anyone have any pointers to source code to generate smpte?
that's me :-)
ftp://www.iua.upf.es/pub/mdeboer/projects/SMPTE/
you can find the sourcecode for the SMPTE decoder, and some reallife
SMPTE signals (from analog tape)
i just found the whole smpte discussion in the archives. it's a nice
read, complete forgot that it was such a long thread. it starts with
http://eca.cx/lad/2000/Jul/0287.html
by the way, i quote from one of the messages:
i wish to god that SMPTE had never taken off. its the most ridiculous
On 19 Nov 2002 15:29:19 -0500
jfm3 [EMAIL PROTECTED] wrote:
I'm running the low latency and preemption kernel patches on 2.4.19. I'm
getting disk copy latency of around 10ms. Since I'm a real time person
(according to M. Goggins) who doesn't use the hd in real time, I can
live with this, but
On my asus A7V266-E board i had to set down dma mode from DMA100 to DMA33
get LL working fine. (3ms)
That is very interesting information (I have the same board). Did you discover
this by trial and error?
Maarten
I have the same problem as you. And the additional problem
of having the PC in the livingroom, where I am torturing my
daughter and girlfriend with the noise :-(
Anyway, when I bought my pc, the already gave me a
more expensive fan, because the got so many people
complaining about the noise.
Really? I've got an A7V266E and haven't noticed anything hot.. and
anyway, I've never heard of a motherboard making a system hot in the
first place.
I have no prove of this, it is just something I heard. But I do have
to say mainboard temp gets high, but I am not sure if this is the
Where are you reading your temperatures? You sure that it's the
motherboard and not the CPU? Do you know where on the motherboard this
reading is coming from (the chipset, or just a general ambient case
thermometer)?
In the BIOS. I suppose it is the chipset.
What type of climate do
Hello,
Storing large arrays of floats in XML files is very
unefficient when it comes to space. Where in a binary
format each float would contain only 4 bytes, in a
ASCII format this number much bigger.
A simple solution would be to store the values as
binary inside the XML. It is too bad XML
So, I was thinking to use binhex, or something similar,
encoding the 32 bits in 7 bit values in the ascii
character range. Thus, each float would take 5 bytes.
i am wrong.. binhex, uuencode, xxencode and base64 use
6 bit values, also known as 3-in-4 encoding (3 8-bit
values = 24 bit - 4 6-bit
the thing i like about xml is it's human-readable.
doing binary in xml combines all the bloat of the former with the
ugliness of the latter...
yes...
do you expect to have such masses of float data that size will become an
issue ?
yes. lots of FFT bins, sinusoidal peaks and tracks... a 5
I take this to linux-audio-dev...
Feature request: alsa 0.9 support, preferably including
alsa sequencer support. I want to play these synths from
If I can help please let me know ! What about MIDI control of the
potentiometer / slider / switches ? I own all kind of doepfer's MIDI
On Fri, 26 Apr 2002 00:37:21 +0200 (CEST)
[EMAIL PROTECTED] wrote:
Hello,
I am just working (finishing) on modular realtime effect procesor. It
is my thesis project. I would like to ask three questions:
1.
How can I calculate the latency in ms with OSS driver?
I would strongly advice you
Hello,
I just updated alsamixergui and aconnectgui. The code is now up to date
with alsa-utils 0.9.0rc1.
alsamixergui and aconnectgui are fltk front-ends for alsamixer
and aconnect. thet are written directly on top of the alsamixer and
aconnect source, leaving the original source intact, only
Hello Patrick,
Looks very nice.
Am I correctly assuming that you will be the maintainer of the ALSA pages
from now on?
Could you add the following alsa 0.9 applications I wrote:
http://www.iua.upf.es/~mdeboer/projects/tapiir/
http://www.iua.upf.es/~mdeboer/projects/aconnectgui/
i have been thinking about the following before, and now
that i read about it on slashdot, and it got me thinking
a bit more.
the issue is sound cancelation. the article mentioned
on slashdot,
http://www.newscientist.com/news/news.jsp?id=ns2094
talks about a (very expensive) general approach
Hello,
I sent the following to the ALSA list, but did not get any answer, so I
repost it here, hoping anybody has delt with this before.
First, I noticed that the alsa rtctimer is not compiled when you have
the RTC in your kernel as a module. The problem is in
alsa-kernel/core/Makefile, where
Hello,
I mainly write this mail to Paul Davis, but of course everybody is
welcome to answer :-)
I am trying to do realtime audio i/o in combination with disk input.
I know that Paul has a lot of experience with this. Rather than
looking through his code, I'd appreciate some description of the
Thanks. Looking through Mustajuuri code, I have the impression you use
the same approach as I do (that's at least reassuring). I will do some
tests to see if the same problems occur, which would mean that they are
caused somewhere else.
Maarten
Thanks Paul for your reply.
don't ever run application threads at max priority. it makes it
impossible to construct a watchdog for runaways.
Okay, I'll make that max-1
you don't need to use mutexes for the buffer itself. when there is one
reader and one writer, you can use a lock free
aconnectgui is a FLTK based frontend for aconnect. It is written
directly on top of the aconnect source, leaving the original source
intact, only adding a couple of ifdefs, and some calls to the gui
part, so it provides exactly the same functionality, but with a
graphical userinterface.
yes, that sounds like a layout problem. i will try to get my hands on
a SBLive, and also use some virtual midi ports. as you might have noticed,
this release is still rather beta: lot's of printf to be removed ;-)
maarten
I just completed version 01 of a very scrappy but functional gui midi
patch bay for alsa's seq api It's essentially just a graphical version
of aconnect It's available from:
I wrote one already a long time ago (though for alsa 05x, but it should
be easy to port)
On Tue, 05 Mar 2002 08:33:53 -0500
Dave Phillips [EMAIL PROTECTED] wrote:
Maarten de Boer wrote:
...my alsamixergui (which I have a version for
0.9.x., based on the latest CVS sources, I should make it available A.S.A.P)
Please !!!
You can download it from
http://www.iua.upf.es
Hello,
Having changed my soundcard (from Trident 4D Wave NX to ens1371)
while the trident driver is broken, I am now achieving good
low latency results on my AMD Athlon, with a kernel compiled
for Athlon. I moved to the latest kernel, 2.4.16, with Andrew
Morton's low latency patch, but when I
Andre Pang [EMAIL PROTECTED] wrote:
I am thinking of setting up a webpage, where people can post
their latencytest results, so we can keep an inventory of
the several combinations, categorising on:
This is a great idea. I actually started re-writing the latency
testing harness scripts
Takashi Iwai [EMAIL PROTECTED] wrote:
Hmm, I'm using an Athlon 600 MHz and got promising results even with
it. You can find the results (on different conditions) under
http://alsa-project.org/~iwai/latency-results/2.4.12
The result with all LL-patches is found in ll-all directory.
On Thu, 15 Nov 2001 18:35:33 +0100
Andy Lo-A-Foe [EMAIL PROTECTED] wrote:
On Thu, Nov 15, 2001 at 05:28:19PM +0100, Maarten de Boer wrote:
It would be really nice if somebody could repeat my tests on identical or
similar hardware (AMD Athlon, Trident 4DWave NX), with the same versions
Or, will you try AA patches? It's not a bad idea, since the current
(vanilla) VM is based on Andrea's code. Linus still doesn't include
all his patches.
Yes, I will try the AA patches, though I am not very sure which to use,
and on which kernel. The most recent one on a standard kernel
Okay, here's my setup:
- kernel 2.4.15-pre4 + Robert Love's preempt patches
(http://www.tech9.net/rml/linux/)
- kernel HZ value set to 1000, default is 100
(see /usr/src/linux/include/asm/param.h)
- alsa 0.9.0beta9 + Trident 4DWave NX
- SCSI hard disk with ReiserFS
- Geforce2 GTS
On Thu, 15 Nov 2001 13:50:39 -0500
Paul Davis [EMAIL PROTECTED] wrote:
Hmm sounds like h/w problem, then..
It's interesting to know where the data is contaminated, whether on
the driver level or transfer between capture and playback on
user-space, or what else..
I would say on driver
When I compile my kernel for i386 on an AMD Athlon, the
alsa latency test is giving me weird distortion (knispering
sound, like a dusty vinyl record)
$ arecord -r 44100 -f S16_LE | aplay
does not seem to have this problem.
I investigated this matter a bit further, and the crackling sounds
I've used this setting:
echo 6 /proc/sys/vm/vm_scan_ratio
echo 2 /proc/sys/vm/vm_mapped_ratio
echo 4 /proc/sys/vm/vm_balance_ratio
Uuhh... I don't have these files...
$ ls /proc/sys/vm/
bdflush kswapd overcommit_memory page-cluster pagetable_cache
Did I miss something in my kernel
On Tue, 13 Nov 2001 14:40:42 -0600
Christopher Deckard [EMAIL PROTECTED] wrote:
i'm getting ~8ms on an AMD K6 running 2.2.17 LL under RH6.1, and using
uncsound 4.x.
that's not 3ms either...
Maarten
I did not get any reaction on one of my previous mails, so I send it
again. There have been numerous reports on this list about good LL,
and now I find it very frustrating I can not get it myself. Please
help me out! I'd really hate the get the impression that the whole
LL issue is still unreal.
Hello,
In reply to D. Stimits and Roger Larsson about compiling
an LL kernel for AMD Athlon without the 3DNOW optimizations
I still get the undefined references to the
*_mmx_* functions. I am perfectly sure that I really recompiled
it all, and to demonstrate it, here follow the steps I toke:
Hello,
When trying to run jack, I got the following error:
jackd: error in loading shared libraries: jackd: undefined symbol: mknod
This was easily solved adding #include unistd.h in engine.c
Also, to compile the fltk client, I had to change the order of
arguments: move the object file to
I made a webpage with the several tests I did. Please have a look
at it. I ran tune_disk /dev/hda, and
do_tests none 3 256 0 2048
(indeed a really small size for the disk i/o test, but I was more
interested in the rest)
http://193.145.55.36/latencytest
The soundcard I use is a Trident 4DWave
I now try to compile my kernel for AMD K7, but without 3DNow disabled.
Check if you have this in .config
CONFIG_X86_USE_3DNOW=y
I remove this line, and get the following undefined references:
ld -m elf_i386 -T /mnt/hda8/src/linux-2.4.14-LL/arch/i386/vmlinux.lds -e stext
Hi, I wrote this before, but I did not get any response.
When I compile my kernel for i386 on an AMD Athlon, the
alsa latency test is giving me weird distortion (knispering
sound, like a dusty vinyl record)
$ arecord -r 44100 -f S16_LE | aplay
does not seem to have this problem.
It really does
Since with AMD Athlon I get either bad low latency (compiling for AMD)
or bad sound quality (if I compile for 386. really don't understand why..),
I now did some tests on a Pentium III. The results are still not good:
kernel 2.4.14 low latency patch a Pentium III 800 Mhz,
dave willis [EMAIL PROTECTED] wrote:
my system is an amd athlon 900 mhz [...]
Dave here shows his LL results. However,
Roger Larsson wrote:
But wait... an AMD
They have some 3DNow specific kernel optimizations that causes trouble -
they need to disable preemtion (uses fp registers...)
Not understanding why I could not get alsa-lib/test/latency.c to work
well, I decided to run Benno's latency test. It has been a long time
since I did... The results are surprising, and nothing like the excellent
results I have seen posted here.
$ ./do_tests none 3 128 0 256
Abramo Bagnara [EMAIL PROTECTED] wrote:
This would not imply a major rewrite: the API and the kernel code has
been thought for that.
You need simply to have different snd_pcm_tick_set functions (see
pcm_lib.c).
That's interesting. So you think that that would be the prefered way
to do low
Hello,
I am using alsa 0.9.0beta8a, and I am having big difficulties
with synchronous I/O. I am using the latency.c test. When I
use it as it is, it eats all the CPU for 30 seconds (well, 3,
I changed to value the second time I tried). I can understand
why this happens, but it is certainly not
Hello,
As you may have heard, SEGA discontinued the Dreamcast,
and you can buy them really cheap now. Looking at the
specs, I realized that the Dreamcast would be a great
standalone box for audio applications. And it runs Linux,
or the open source OS KallistiOS. The CPU of the Dreamcast
is a
[...]
I got your reference from the Nigeria Exports Promotion Council (NEPC)
[...]
Okay guys, confess: who gave the mailinglist address to the NEPC?
1 - 100 of 102 matches
Mail list logo
