You are here

Unit test fails (SpanFileIOTests)

4 posts / 0 new
Last post
Unit test fails (SpanFileIOTests)


I try to install the latest release of Rosetta (Saturday, May 17, 2014). I have compiled the debug mode of Rosetta with succes, but I when I try to run the Unit test I have this error on the consol :

-------- Unit test summary --------
Total number of tests: 1527
number tests passed: 1520
number tests failed: 7
failed tests:
core.test: SpanFileIOTests:test_emptyInput
core.test: SpanFileIOTests:test_helixOutofBounds
core.test: SpanFileIOTests:test_singleHelix
core.test: SpanFileIOTests:test_negativeTotalRes
core.test: SpanFileIOTests:test_multipleHelix
core.test: SpanFileIOTests:test_fullHelix
core.test: SpanFileIOTests:test_edgeHelix
Success rate: 99%
---------- End of Unit test summary

I have tried to recompile another time this release of Rosetta and the previous one but I still have this seven tests which fails.

Francois Inizan

Post Situation: 
Sat, 2014-06-07 10:10

It's a little hard to debug with just the summary. I'd recommend re-running just the SpanFileIOTests suite, with output on (remove "--mute all" from your test/ script if you used it, and add "-1 SpanFileIOTests"). Seeing the output of that command would greatly help with debugging.

The other option is just to ignore it, especially if you're not working with membrane proteins. The SpanFileIO tests are somewhat recent additions, and the code it tests is only used for working with membrane proteins. The test passed on our test server for that revision (, assuming you're using weekly version 2014.20.56838), and it doesn't look like other tests are having problems. It could be some numeric precision/compiler issues on your machine that makes a (spurious) change in the test result. What compiler and platform are you using?

Mon, 2014-06-09 10:29

Hello, thank you for your reply,

I re-run the SpanFileIOTest, this is the result :

Identifying platform...

Platform found: debug/linux/3.13/64/x86/gcc/4.8/default
Command line:: cd build/test/debug/linux/3.13/64/x86/gcc/4.8/default && ./core.test SpanFileIOTests --database /home/frini/Programms/ROSETTA/rosetta_2014.20.56838_bundle/main/database
Running one suite: SpanFileIOTests
Test suite: SpanFileIOTests (test/core/membrane/io/SpanFileIO.cxxtest.hh)
core.init: Rosetta version from
core.init: command: ./core.test --database /home/frini/Programms/ROSETTA/rosetta_2014.20.56838_bundle/main/database
core.init: 'RNG device' seed mode, using '/dev/urandom', seed=-122285480 seed_offset=0 real_seed=-122285480
core.init.random: RandomGenerator:init: Normal mode, seed=-122285480 RG_type=mt19937
core.init.random: RandomGenerator:init: _RND_TestRun_ mode, seed=1000 RG_type=mt19937
Trace: Testing response to empty or null input...
core.test: src/ObjexxFCL/DynamicIndexRange.hh:103: ObjexxFCL::DynamicIndexRange::DynamicIndexRange(int): Assertion `legal_static()' failed. Building topology object from spanfile

I'm using gcc-4.8 and g++-4.8 on the weekly version 2014.20.56838. with Xubuntu 14.04.
In my case, I want to run for an Ab Initio prediction, is it possible ?

Thanks again for your help !
François Inizan

Mon, 2014-06-09 14:10

So the test that is failing is one to test parsing of malformed span files. Reading the code, it looks like we're currently lacking enough good checks for error conditions, so depending on how things run, it's very possible that this test might fail on a your system.

If you're not using membrane proteins, I wouldn't worry about it. Even if you are using membrane proteins, as long as you have a well-formatted spanfile you should be okay.

Tue, 2014-06-10 11:25