Time Series Schema Issues

[Arnold Rots]

Conversation at lunch was somewhat difficult, so I thought it might be
good to summarize my understanding of what John Brewer and I discussed
and what the remaining issues are.

I am assuming that we are considering a table wit han STC header, such
as I presented and with a body structure similar to John's.

I see three issues that remain to be decided:

1 Assuming that we allow multiple tables, will we allow each table to
contain only one quantity, in addition to timestamps, or a limited
number of named columns? In a sense, the former presents a transposed
version of the complete table: each column is provided, more or less,
as a row. Allowing multiple (6, 10, 12, 20?) columns would allow
people the option to put small tables in traditional table form.

2 We need to allow different data types. This can be done by
enumerating named columns of different data types in the schema and
let the document choose from them, or by deriving different types from
the same parent type and use "xsi:type=" in the document.

3 It would be nice if errors and confidence intervals (or: asymmetric
errors) could be brought to he same level. At this point, a simple
error (sigma) is a column by itself, while upper and lower error
limits are attributes of the value. One should be able to select
either a simple error or upper and/or lower limit.

  - Arnold

--------------------------------------------------------------------------
Arnold H. Rots                                Chandra X-ray Science Center
Smithsonian Astrophysical Observatory                tel:  +1 617 496 7701
60 Garden Street, MS 67                              fax:  +1 617 495 7356
Cambridge, MA 02138                             arots at head.cfa.harvard.edu
USA                                     http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------