Bug in image rendering of dative bonds

[IvanChernyshov]

In some cases, dative bonds are painted in the wrong color of the end atom:

Chem.MolFromSmiles('N->1CCN->[Cu++]1')

or are even invisible, while the molecular graph is correct:

mol = Chem.MolFromSmiles('[Cu++]->1->2.N1CCN2')
mol



mol.Debug()
RDKit INFO: Atoms:
RDKit INFO: 	0 29 Cu chg: 2  deg: 2 exp: 0 imp: 0 hyb: 6 arom?: 0 chi: 0
RDKit INFO: 	1 7 N chg: 0  deg: 2 exp: 2 imp: 1 hyb: 4 arom?: 0 chi: 0
RDKit INFO: 	2 6 C chg: 0  deg: 2 exp: 2 imp: 2 hyb: 4 arom?: 0 chi: 0
RDKit INFO: 	3 6 C chg: 0  deg: 2 exp: 2 imp: 2 hyb: 4 arom?: 0 chi: 0
RDKit INFO: 	4 7 N chg: 0  deg: 2 exp: 2 imp: 1 hyb: 4 arom?: 0 chi: 0
RDKit INFO: Bonds:
RDKit INFO: 	0 1->2 order: 1 conj?: 0 aromatic?: 0
RDKit INFO: 	1 2->3 order: 1 conj?: 0 aromatic?: 0
RDKit INFO: 	2 3->4 order: 1 conj?: 0 aromatic?: 0
RDKit INFO: 	3 0->1 order: 17 conj?: 0 aromatic?: 0
RDKit INFO: 	4 0->4 order: 17 conj?: 0 aromatic?: 0

Unfortunately, I was not able to find any specific pattern of this bug.

[DavidACosgrove]

I'll have a look at this. It was my fault, more than likely. I don't think the dative bonds split the bond into 2 different colours depending on the ends, and it looks like it doesn't set the draw colour to anything sensible, just uses whatever it was using last. If the dative bond is the first one it draws, then the colour will be the background colour.

[DavidACosgrove]

Ivan, can you say which version of RDKIt you're using, please? This is something I thought I'd fixed in 2020.03.1 (at least, the missing bonds bit) and I'm not reproducing the problem in the new version.

[greglandrum]

Closed by #3066