[DAS2] Notes from DAS/2 code sprint #3, day five, 18 Aug 2006

[Andrew Dalke]

> Action items are flagged with '[A]'.

I see there aren't any.

> ad: need to mention to lincoln and berkeley folks. specify what the
> algorithm is to


I would have added this as an [A].

> Post-teleconference Discussion
> -------------------------------
>
> gh: would you be willing to give up multiple locations in the spec?
> aday: would you be willing to give up bidirectional parent-child 
> pointers?
> gh: let me think about it...

Regarding bidi pointers (btw, should we change the tag <PART> to 
<CHILD>?)

As someone parsing GFF3 it's annoying that I have to keep any
features with an ID hanging around until the end just in case someone
wants to refer to it later as a parent.  GFF3 does have a directive
to allow flushing but so far I haven't come across a gff3 file which
uses it.  Having bidi links makes this trivial.

As someone concerned about database integrity issue, what happens
with a writeback which says "I am a child of X and Y" where X and Y
were not previously connected.  (Or "I am a parent of...", depending on
the link direction.)  Does the server allow that?  Reject that?


Regarding multiple locations, it's a data modeling issue.  If a
record Q has N multiple locations then it's identical to a record
Q' with N children, Q1', Q2', ..., Qn'.  The Qi' will have a different
type record than Q'.

As a plus, or minus, each Qi' will be annotatable, have it's own
identifier, etc. which is not the case if features can have multiple
locations.

					Andrew
					dalke at dalkescientific.com