[DAS2] Fwd: DAS writeback

Andrew Dalke dalke at dalkescientific.com
Tue May 2 14:43:09 UTC 2006


I emailed Nomi regarding talking about writeback.  She suggested
I talk with Mark Gibson instead.  Here's the core part of his
response on splits&joins

[Andrew]
>> One option is to track exactly what happened.  For exampe, if
>> features F1 and F2 are merged into F3 then either
>>   DELETE F1
>>   DELETE F2
>>   CREATE F3
>>      <derives from F1>
>>      <derives from F2>
>> -or-
>>   DELETE F1
>>   MODIFY F2
>>      -change the range-
>>      <derives from F1>

[Mark]
> this is essentially what apollo does and i dont like it.
>
> in apollo the user chooses the name from a merge which can be F1-name, 
> F2-name or temp-name (new name from db) - if F1 is picked then f2 
> should be deleted & vice versa - apollo doesnt do this proper at the 
> moment (historical reasons - user didnt use to pick the name & the 
> name picking was hacked yada yada...)

[Andrew]
>> In this case the <derives from *> means the information is included
>> in the feature record.
>> I proposed an alternative, which I'm calling a microdelta.
>>   REASON: merge
>>   DELETE F1
>>   DELETE F2
>>   CREATE F3

[Mark]
> i like this one the best. I think its valuable to record in the record 
> that merge (&splits) have occured - and this makes it the most 
> obvious. Some databases like to track merge & splits (like TAIR - i 
> dont think chado tracks this but it should)
>
> i think its worthy on its own and its muddled to shove this in the 
> feature with derives from.


					Andrew
					dalke at dalkescientific.com




More information about the DAS2 mailing list