Patrick and All,
I figured out my original error. In my XML file I was trying to use a variable without including it in the line:<make>ACK.Text_Sanitize($msg)</make>
Logan Washbourne
Electrical Engineering Graduate StudentOn Sat, Aug 22, 2015 at 1:57 PM, Washbourne, Logan <lwashbo@ostatemail.okstate.edu> wrote:
I appreciate the advice, I've been pretty diligent about using sudo ldconfig after I read that not using it could create problems. All advice helps!Richard,So the RuntimeError seems to be the problem, I'm leaning towards the idea that I am not handling the PMT's correctly. Thoughts?My XML file can be found here: https://gist.github.com/loganwashbourne/68367a93b7fe9bf28bceWhen I run "python top_block.py" this error occurs:the new line looks like this:self.ACK_Text_Sanitize_0 = ACK.Text_Sanitize("String").Patrick,Thanks for the advice! So I added that string parameter(or char array) to the Top_Block.py file
comm1@comm1:~/Logan/Thesis$ python top_block.py
Traceback (most recent call last):
File "top_block.py", line 92, in <module>
tb = top_block()
File "top_block.py", line 65, in __init__
self.ACK_Text_Sanitize_0 = ACK.Text_Sanitize("String")
File "/usr/local/lib/python2.7/dist-packages/ACK/ACK_swig.py", line 399, in make
return _ACK_swig.Text_Sanitize_make(*args, **kwargs)
RuntimeError: attempt to set_msg_handler() on bad input message port!(Electromagnetics)Logan WashbourneElectrical Engineering Graduate StudentOn Fri, Aug 21, 2015 at 4:37 PM, Patrick Sathyanathan <wpats@hotmail.com> wrote:I see the following in the output of "nm -C -u":U gr::ACK::Text_Sanitize_impl::forecast(int, std::vector<int, std::allocator<int> >&)This was the undefined symbol that was causing the module import to fail... as you have discovered yourself. Now that the module import has succeeded you are seeing a different error.This error is because of the following in the generated python file:self.ACK_Text_Sanitize_0 = ACK.Text_Sanitize()Note that this will invoke "Text_Sanitize::make" which expects a "char * message" argument. That causes the error message below. For some reason GRC is not adding that parameter in the above statement. Did you add a declaration for that parameter in the XML file ?To verify that the XML file is the issue just try editing the generated python file and changing the above to:self.ACK_Text_Sanitize_0 = ACK.Text_Sanitize("some string")and running it from the command line (run "python top_block.py" in a terminal window).--Patrick
Date: Thu, 20 Aug 2015 11:08:56 -0500
From: lwashbo@ostatemail.okstate.edu
To: discuss-gnuradio@gnu.org
Subject: Re: [Discuss-gnuradio] OOT Module Attribute Error module object has no attribute 'blockname'I do apologize for the long questions, if any of you feel like I need to spend more time looking into this myself before asking the mailing list, please don't hesitate to mention it.The new error is :Last thing, I realized that I am creating the forecast function in the Text_Sanitize_impl.h file but not referencing it it the .cc file (I commented it out). I tried commenting out the void deceleration of the forecast function in the .h file but then I get a new error when I try to run the grc file(which is just a constant int source connected to my Text_Sanitize block, which is connected the the message debug "print" port).Next, the ACK_swig.i file can be found here :I really don't think I have any callbacks in my XML code, I just reference certain input variables in the XML code.I'll be honest, I'm not really sure what I'm looking at in this return, I do see some error statements but I'm not sure if they are just stating how it would handle an error or if it is an actual error.and the results of that can be found here: https://gist.github.com/loganwashbourne under the nm -C -u libgnuradio-<modulename>.so file. (I think github thinks I'm a robot so I can't link to the direct page yet).I started with running: nm -C -u libgnuradio-<modulename>.soNathan and Patrick,Thanks for the tips!
https://gist.github.com/loganwashbourne under the same file name. I checked it against the gr-tutorial swig file and the only difference was that the ACK_swig.i file included a magic2 function call for each of my OOT blocks(check and Text_Sanitize), while the gr-tutorial didn't.
Traceback (most recent call last):
File "/home/comm1/Logan/Thesis/top_block.py", line 92, in <module>
tb = top_block()
File "/home/comm1/Logan/Thesis/top_block.py", line 65, in __init__
self.ACK_Text_Sanitize_0 = ACK.Text_Sanitize()
File "/usr/local/lib/python2.7/dist-packages/ACK/ACK_swig.py", line 399, in make
return _ACK_swig.Text_Sanitize_make(*args, **kwargs)
TypeError: Required argument 'message' (pos 1) not found(Electromagnetics)Logan WashbourneElectrical Engineering Graduate StudentOn Wed, Aug 19, 2015 at 5:06 PM, Patrick Sathyanathan <wpats@hotmail.com> wrote:Hi Logan,I have faced the same error twice recently in my OOT module and finally tracked it down to two causes. The basic reason for the 'object has no attribute' error is that Python's import of your module failed. In the two cases that I saw it was due to undefined symbols in the shared library for my module. My module was called 'tutorial' so the corresponding library is libgnuradio-tutorial.so and you can find it in the .../build/lib directory. To find out the undefined symbols do the following:nm -C -u libgnuradio-<modulename>.soand search for method names in your class. So if your class is 'MyClass' you could do:nm -C -u libgnuradio-<modulename>.so | grep MyClassThe two cases that I ran into were:1. A non-virtual method in my implementation class was undefined because I had forgotten to prefix the method definition with the class name. So the .cc file had "void foo(...)" instead of "void MyClass_impl::foo(...)" and foo was compiled as a ordinary function. Then the command above will report "MyClass_impl::foo" as undefined.2. The second case I ran into was with defining callbacks in the XML file. If you do then you will need to add matching virtual function declarations in the header file. This is needed for SWIG to work correctly. For my case the callback "foo" needed a virtual function declaration "virtual <sometype> foo(...) = 0;" in the actual class declaration (not the ..._impl version) in the header file.Do you have callbacks ? If this is the issue then you should do a "make clean" before running make again to force SWIG to run.Hope this helps.--Patrick
Date: Wed, 19 Aug 2015 14:51:27 -0400
From: natw@ostatemail.okstate.edu
To: lwashbo@ostatemail.okstate.edu
CC: discuss-gnuradio@gnu.org
Subject: Re: [Discuss-gnuradio] OOT Module Attribute Error module object has no attribute 'blockname'My gut is telling me this is a swig problem. I don't know that it's frowned upon, but it's not easy to read without some kind of highlighting that we'd get from github or a gist with files. If I'm correct we'd also need to see swig/ACK.i (probably missing an include and/or gr swig block magic. compare to tutorial swig for sanity check)On Wed, Aug 19, 2015 at 10:38 AM, Washbourne, Logan <lwashbo@ostatemail.okstate.edu> wrote:I looked on the mailing list for that last line error and it pointed me to doing what I mentioned above with the CMakeLists.txt file, but could it be an actual problem with the top_block.py file?The full error thrown is this:to my top level CMakeLists.txt file, because I am using PMT objects in my block, but that didn't get rid of the error.set(GR_REQUIRED_COMPONENTS RUNTIME PMT)I scoured through this webpage(http://gnuradio.org/redmine/projects/gnuradio/wiki/OutOfTreeModulesConfig) and tried adding:Hello all,I know this question has been asked before, several times, but I didn't find a solution that allowed me to use my OOT blocks without running into the error stated in the subject of this email.
Executing: "/home/comm1/Logan/Thesis/top_block.py"
Traceback (most recent call last):
File "/home/comm1/Logan/Thesis/top_block.py", line 92, in <module>
tb = top_block()
File "/home/comm1/Logan/Thesis/top_block.py", line 65, in __init__
self.ACK_Text_Sanitize_0 = ACK.Text_Sanitize()
AttributeError: 'module' object has no attribute 'Text_Sanitize'In the addendum is all of the files I could think would be necessary for someone to look at if they chose to, if including this much text is frowned upon, please let me know.Addendum:
Text_Sanitize_impl.cc
*****************************************************************************
#ifdef HAVE_CONFIG_H
#include "config.h"
Subscribe to: Post Comments (Atom)
No comments:
Post a Comment