From David Adams
To David Alexander
September 9, 2003


David:

Thanks for your comments. I am glad to see that you almost agree with set 
of properties identified in the dataset paper. It suggests we are on the 
right track. It is unfortunate that you disagree with the names because I 
believe naming is very important.

I agree that location and mapping are closely related. One can also throw 
content into the mix because the mapping is from content to location. 
However I believe it is useful to separately identify content and location 
even if they are repeated in the mapping. The mapping may be a very 
complicated and is not needed for all tasks. After splitting, a scheduler 
trying to choose a compute node to process a dataset needs the data 
location but not the mapping.

You specify which properties are essential. I also note that although 
location is essential for logical datasets, it is not essential for 
virtual datasets. A similar comment applies for history.

For naming, my aim is to choose names that are easy to remember and thus 
succinct and descriptive. I prefer to use a single word or an acronym if 
no word can be found. I don't think that the phrases you suggest add enough 
semantic content to justify violating this rule or abbreviating them to 
acronyms.

da