[ensembl-dev] How to handle missing DAS for Ensembl species

[Dan Sheppard]

Hi Ben,

Sorry again for the delay getting back to you. I've fixed that on our 
master branch which will make it into the next release (e76). If you 
want to git cherry-pick it in the meantime, it's got hash 7c590d32...

You will need to regenerate the packeds after applying the fix.

In the process, I discovered a number of races in our startup code for 
generating packeds (which I've not had time to fix yet).

In particular I found, as you may well do, that I needed to delete the 
species and MULTI packeds, as well as config.packed, and restart -r ; 
then delete config.packed (alone) again, and restart -r again. Once they 
are correct there is no further racing during restarts.

As you were interested in various solutions: in the end, I went for the 
merge to be a no-op if the rhs tree is undef.

I don't think it would ever be intuitive, -- as we had, -- to have an 
API to a merging function where the additional data being undef would 
clear existing data, and can't believe that was deliberate (and 
certainly was never deliberately called with such intent).

So getting merge in this case to be a no-op instead, as well as fixing 
the bug, only extends rather than modifies the defined semantics for the 
function.

Do let us know if this doesn't work for you, or you have any further 
suggestions.

Dan.