GNU Radio 3.2 release candidate 2 is now available for download and
testing:
http://gnuradio.org/releases/gnuradio/gnuradio-3.2rc2.tar.gz
http://gnuradio.org/releases/gnuradio/gr-howto-write-a-block-3.2rc2.tar.gz
The associated firmware and FPGA bitstream images for the USRP2 are:
http://gnuradio.org/releases/usrp2-bin/release/txrx_edk10.1_r3.2rc2.bin
http://gnuradio.org/releases/usrp2-bin/release/u2_rev3_ise10.1sp3_r3.2rc2.bin
This release contains all the features, fixes, and bugs of the GNU
Radio development trunk as of 4/14/09.
Our goal is for this to be the last release candidate before the
formal 3.2 release. Those of you with your own GNU Radio projects
who wish to stabilize on the 3.2.x API should begin using the above
instead of the development trunk. We will be making changes to the
trunk software that will break API compatibility soon after the formal
release.
As always, we will maintain the 3.2.x series of releases such that
upgrades to newer releases will not break existing code. In addition,
to the extent possible, we will merge and release selected
functionality from the development trunk that is consistent with this
requirement.
Johnathan C.
Corgan Enterprises LLC
On Tue, Apr 14, 2009 at 3:23 PM, Johnathan C.
[email protected] wrote:
This release contains all the features, fixes, and bugs of the GNU
upgrades to newer releases will not break existing code. In addition,
to the extent possible, we will merge and release selected
functionality from the development trunk that is consistent with this
requirement.
Johnathan C.
Corgan Enterprises LLC
Here is my setup:
OS: Ubuntu 8.04
USRP1 with 2 LFRX and 2LFTX daughterboards
I primary use the USRP as a 4 channel Digitizer and have been running
gnuradio-3.1.3 without any problems so far. To test 3.2rc2 I tried
running the multi_fft.py which is available in the
gnuradio-examples/python/multi-antenna/ directory. I commented out the
portion which checks for the the number of channels available and the
daughterboard ID since I didn’t have the BASIC_RX. Here is the major
change that I found.
The value of len(self.subdev) = 4 in 3.1.3 and =6 in 3.2rc2.
Shouldn’t this be 4 as it was previously?
Regards,
Karthik
This release compiles, checks, installs, and uninstalls properly on
OSX 10.5.6 / XCode 3.1.2 using Apple’s GCC 4.2 and MacPorts (latest
everything, since there’s not much choice), using both “normal” and
VPATH builds. Other pieces of good news:
-
Looks like the GNU Radio build system and programming is compatible
with the just released Boost 1.38 and SWIG 1.3.39.
-
When I upgraded from SWIG 1.3.38 to 1.3.39, the build system worked
perfectly: it recognized that the previous SWIG system files weren’t
available, and recompiled just the SWIG-generated files as well as
regenerated the SWIG dependency file. Much better than erroring out.
On Fri, Apr 17, 2009 at 07:04:25PM -0700, Karthik wrote:
http://gnuradio.org/releases/usrp2-bin/release/u2_rev3_ise10.1sp3_r3.2rc2.bin
OS: Ubuntu 8.04
The value of len(self.subdev) = 4 in 3.1.3 and =6 in 3.2rc2.
Shouldn’t this be 4 as it was previously?
Thanks for pointing this out.
Fixed in [10877] to work with old and new handling of Basic Rx d’boards.
Eric
On Fri, Apr 17, 2009 at 7:04 PM, Karthik [email protected] wrote:
The value of len(self.subdev) = 4 in 3.1.3 and =6 in 3.2rc2.
Shouldn’t this be 4 as it was previously?
It appears this was an inadvertent consequence of the change in the
daughterboard code for the BasicRX and LFRX boards, to allow use as a
complex input device. These daughterboards now have three subdevs
instead of two, so when the above variable is created by concatenating
the daughterboard table from each side, it comes out as six instead of
four.
I can’t test it right now, but I think it is as simple as changing the
comparison to use 6 instead of 4.
Thanks for the bug report.
Johnathan
I checked out the trunk and tried compiling the code. I get the
following error message in in the middle of the compile. I have tried
compiling the latest trunk and also R10877 and still get the same
error. I did the regular ./bootstrap and used
./configure --enable-doxygen --with-boost=/opt/boost_1_36
–enable-usrp2=no --enable-gr-usrp2=no
BTW, you likely don’t need to disable the usrp2 components in the
above, though that isn’t the source of your problem.
g++ -g -O2 -g -O2 -pthread -pthread -Wall -Woverloaded-virtual -o
.libs/test_inband test_inband.o ./.libs/libusrp-inband-qa.so
/home/anankekarthik/gnuradio/usrp/host/lib/inband/.libs/libusrp-inband.so.0:
undefined reference to `usrp_standard_rx::tune(int,
boost::shared_ptr<db_base>, double, usrp_tune_result*)’
This looks like the build system is getting confused as to the
location of libusrp (which contains the functions above). Just to
start from a known configuration, can you do a ‘sudo make uninstall’
to ensure the system locations are devoid of any GNU Radio
libraries/header files, the start from scratch with the bootstrap
step?
Johnathan
On Fri, Apr 17, 2009 at 7:25 PM, Eric B. [email protected] wrote:
http://gnuradio.org/releases/usrp2-bin/release/txrx_edk10.1_r3.2rc2.bin
release.
Here is my setup:
The value of len(self.subdev) = 4 in 3.1.3 and =6 in 3.2rc2.
Shouldn’t this be 4 as it was previously?
Thanks for pointing this out.
Fixed in [10877] to work with old and new handling of Basic Rx d’boards.
Eric
I checked out the trunk and tried compiling the code. I get the
following error message in in the middle of the compile. I have tried
compiling the latest trunk and also R10877 and still get the same
error. I did the regular ./bootstrap and used
./configure --enable-doxygen --with-boost=/opt/boost_1_36
–enable-usrp2=no --enable-gr-usrp2=no
g++ -g -O2 -g -O2 -pthread -pthread -Wall -Woverloaded-virtual -o
.libs/test_inband test_inband.o ./.libs/libusrp-inband-qa.so
/home/anankekarthik/gnuradio/usrp/host/lib/inband/.libs/libusrp-inband.so.0:
undefined reference to usrp_standard_rx::tune(int, boost::shared_ptr<db_base>, double, usrp_tune_result*)' /home/anankekarthik/gnuradio/usrp/host/lib/inband/.libs/libusrp-inband.so.0: undefined reference to usrp_standard_tx::tune(int,
boost::shared_ptr<db_base>, double, usrp_tune_result*)’
/home/anankekarthik/gnuradio/usrp/host/lib/inband/.libs/libusrp-inband.so.0:
undefined reference to
usrp_standard_rx::determine_rx_mux_value(usrp_subdev_spec const&)' /home/anankekarthik/gnuradio/usrp/host/lib/inband/.libs/libusrp-inband.so.0: undefined reference to usrp_standard_tx::determine_tx_mux_value(usrp_subdev_spec const&)’
/home/anankekarthik/gnuradio/usrp/host/lib/inband/.libs/libusrp-inband.so.0:
undefined reference to usrp_basic::selected_subdev(usrp_subdev_spec const&)' /home/anankekarthik/gnuradio/usrp/host/lib/inband/.libs/libusrp-inband.so.0: undefined reference to db_base::dbid()’
Karthik
Johnathan C. wrote:
GNU Radio 3.2 release candidate 2 is now available for download and testing:
http://gnuradio.org/releases/gnuradio/gnuradio-3.2rc2.tar.gz
http://gnuradio.org/releases/gnuradio/gr-howto-write-a-block-3.2rc2.tar.gz
[etc]
My 3.2-rc2 build fails with the following error:
Making all in tests
make[4]: Entering directory
/home/jan/gr-build/gnuradio-3.2rc2/gnuradio-core/src/tests' /bin/bash ../../../libtool --tag=CXX --mode=link g++ -g -O2 -g -O2 -pthread -pthread -Wall -Woverloaded-virtual -o test_all test_all.o ../../../gnuradio-core/src/lib/libgnuradio-core-qa.la /home/jan/gr-build/gnuradio-3.2rc2/gnuradio-core/src/lib/libgnuradio-core.la libtool: link: g++ -g -O2 -g -O2 -pthread -pthread -Wall -Woverloaded-virtual -o .libs/test_all test_all.o ../../../gnuradio-core/src/lib/.libs/libgnuradio-core-qa.so -L/usr/lib /usr/lib/libcppunit.so -ldl /home/jan/gr-build/gnuradio-3.2rc2/gnuradio-core/src/lib/.libs/libgnuradio-core.so /home/jan/gr-build/gnuradio-3.2rc2/omnithread/.libs/libgromnithread.so /home/jan/gr-build/gnuradio-3.2rc2/gruel/src/lib/.libs/libgruel.so -lboost_thread-gcc43-mt-1_35 -lrt /usr/lib/libfftw3f.so -lgsl -lgslcblas -pthread ../../../gnuradio-core/src/lib/.libs/libgnuradio-core-qa.so: undefined reference to CppUnit::assertDoubleEquals(double, double, double,
CppUnit::SourceLine)’
collect2: ld returned 1 exit status
make[4]: *** [test_all] Error 1
make[4]: Leaving directory
/home/jan/gr-build/gnuradio-3.2rc2/gnuradio-core/src/tests' make[3]: *** [all-recursive] Error 1 make[3]: Leaving directory /home/jan/gr-build/gnuradio-3.2rc2/gnuradio-core/src’
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory
/home/jan/gr-build/gnuradio-3.2rc2/gnuradio-core' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory /home/jan/gr-build/gnuradio-3.2rc2’
make: *** [all] Error 2
This is on Ubuntu 9.04, following the instructions on the wiki
(http://gnuradio.org/trac/wiki/UbuntuInstall). cppunit is 1.12.1-1 as
installed from the Ubuntu repositories. Some output from nm:
jan@ubuntu:~$ nm --demangle /usr/lib/libcppunit.a | grep
assertDoubleEquals
00000030 T CppUnit::assertDoubleEquals(double, double, double,
CppUnit::SourceLine, std::string const&)
So there seeems to be an extra parameter in the version of
assertDoubleEquals that I have. According to cppunit documentation,
there are macros with and without the extra parameter (which is a
user-supplied message in case of error).
Thanks,
Jan
On Mon, Apr 20, 2009 at 8:08 AM, Johnathan C.
[email protected] wrote:
libraries/header files, the start from scratch with the bootstrap
step?
Johnathan
Thanks, that worked!
Karthik
Has this specific source code tree been compiled before under a
different environment? Can you do a ‘make distclean’ and try again
(from the ./bootstrap step)?
I was working from the tarball, so this was a completely new source
tree.
However, I think I found the culprit after some more digging. There was
an
older version of cppunit hiding in /usr/local/lib (wasn’t installed
through the package manager, so it escaped my attention for a while).
The
build is still underway, but I am pretty sure this is the root cause.
Sorry about the false alarm!
Cheers,
Jan
On Thu, Apr 30, 2009 at 12:52 AM, Jan S. [email protected]
wrote:
…/…/…/gnuradio-core/src/lib/.libs/libgnuradio-core-qa.so: undefined
reference to `CppUnit::assertDoubleEquals(double, double, double,
CppUnit::SourceLine)’
[…]
jan@ubuntu:~$ nm --demangle /usr/lib/libcppunit.a | grep assertDoubleEquals
00000030 T CppUnit::assertDoubleEquals(double, double, double,
CppUnit::SourceLine, std::string const&)
So there seeems to be an extra parameter in the version of
assertDoubleEquals that I have. According to cppunit documentation, there
are macros with and without the extra parameter (which is a user-supplied
message in case of error).
I can’t replicate your results here on an Ubuntu 9.04 machine.
There is only one actual assertDoubleEquals function in libcppunit,
and it takes all five arguments. The macro
CPPUNIT_ASSERT_DOUBLES_EQUAL that takes three arguments supplies the
other two as defaults. This has been the case on the default
installation of libcppunit for Ubuntu 8.04, 8.10, and 9.04. I’m
trying to figure out how your libgnuradio-core-qa.so came to have a
dynamic link to a version of the function with only four arguments.
Has this specific source code tree been compiled before under a
different environment? Can you do a ‘make distclean’ and try again
(from the ./bootstrap step)?
Johnathan
