Ctest freezes in random moments on my OOT

GNU Radio
1

Hello,

I’ve written a simple block that takes a c34vector message as input and
outputs a message containing average mag^2 of the samples in the vector.

I’ve added the QA code in Python and it seems to be working except the
test
never finishes. I’ve added a few print statements to see whats up. From
what I can gather:

  • The freeze doesn’t use up CPU (at least nothing noticeable)
  • Random number of messages are processed before the freeze occurs,
    sometimes none are sent

Could someone give me a hint on that? I’ve added the relevant code in
the
attachment.

BR,
Michal

2

On Thu, Aug 14, 2014 at 6:38 AM, Michal J. [email protected]
wrote:

sometimes none are sent

Could someone give me a hint on that? I’ve added the relevant code in the
attachment.

BR,
Michal

What version of GNU Radio are you using? There is a known problem with
message-only apps not properly shutting down. This should be fixed in
the
latest 3.7.4 release, but I think 3.7.3 is still buggy on this one.

Tom

3

On Thu, Aug 14, 2014 at 4:15 PM, Tom R. [email protected] wrote:

From what I can gather:

What version of GNU Radio are you using? There is a known problem with
message-only apps not properly shutting down. This should be fixed in the
latest 3.7.4 release, but I think 3.7.3 is still buggy on this one.

Tom

I was on 3.7.2, I believe. I used the build-gnuradio script to update to
3.7.4. My problem persists. However, odd things happen. I doubt it is
relevant to my actual problem, but here it goes:

5:

5: FAIL: test_001_t (main.qa_burst_to_power)
5:

5: Traceback (most recent call last):
5: File
“/home/mehow/projects/gr-gsm_measure/python/qa_burst_to_power.py”, line
124, in test_001_t
5: msg_pow.message_ports_in(), 0)), ‘bursts’)
5: AssertionError: ‘system’ != ‘bursts’
5:
5:

5: Ran 1 test in 0.005s

I don’t have a port called ‘system’. This happened only once and made me
tweak with the assertion statements. Now, when I check the message ports
against their actual names (‘bursts’ and ‘power’), the assertion is
passed,
but when I change the name to check against in the QA code I get (for
both
of my ports):

5: AssertionError: ‘system’ != ‘whatever_name_I_put_there’

Other than this oddity, I get exactly the same thing as at the
beginning.
Could this happen because I have some leftovers of the previous version
of
gnuradio after using the script?