Failing moleculetest - bondByUniqueID

[ghutchis]

The current Travis is failing out for all builds after #336. That merge was tested without #316 so ctest didn't show the failure.

I have no idea why the test is failing qtgui/moleculetest.cpp#L544

  EXPECT_FALSE(assign.bondByUniqueId(2).isValid());

The commit doesn't change anything in the qtgui/molecule class.

Help debugging this would be most welcome (e.g., how many bonds does the water molecule think it has at this point, etc.)

[cryos]

I can take a look at this, I have been traveling a lot, but will try to set some time aside to figure out what is wrong unless @psavery gets a chance to look into it before me.

[psavery]

I believe I found a solution and submitted PR #349 as a fix.

The issue is that QtGui::Molecule::addBond for atom types was appending to m_bondUniqueIds, then calling Core::Molecule::addBond for atom types, which then called QtGui::Molecule::addBond for atom indices because QtGui overrides the Core version of this function, and then m_bondUniqueIds would get appended a second time.

My fix calls Core::Molecule::addBond for atom indices directly, so it circumvents this complex calling sequence.

If another fix is preferred, let me know.

[ghutchis]

Thanks - this is fixed by #349