by TerryG » Sep 18th, '08, 16:50
OK, I think I have got it clear in my head now. The table I will upload will now contain three columns; uid, txt, urlTxt. SDBX should allow my uid column to be used as the item name and I will look after ensuring it's unique for the domain. There will now only be two, additional name-value pairs for each item; txt and one for urlTxt. Each of them will have a limited (likely one) number of attribute values associated with them. So, in the example described in my original post, I would now expect the SimpleDB domain to contain 10 items once the upload is complete. This is close to what SDBX is delivering now, but I would suggest the following;
1) In the event that the attribute name for a name-value pair exists for a given item, give the user the option of either entirely replacing the existing attribute value(s) for that name-value pair, or alternatively, adding the newly uploaded value to the existing value, so long as it's different from the existing value. It should only be necessary to specify this once per upload, as opposed to once-per-item-per-upload.
2) There should be a mechanism which allows a table to be uploaded that contains just item names, and have it delete those items from the domain. This will facilitate keeping the SimpleDB domain in synch with the offline, relational table(s) being use to prepare the upload table in the first place.
Please let me know if you need any additional information or clarification. Thank you very much.