VOEvent session

[Roy Williams]

On 12/10/2010 5:42 AM, Mike Fitzpatrick wrote:
 > My opinions on the added complexity in 2.0 are well-known to the WG
 > leads, but I also promise to document them in the mailing list for
 > the record and yet more discussion.


This is a note on schema complexity for the proposed VOEvent v2.0, 
comparing it with VOEvent v1.1. The cojmplexity is in fact REDUCED in v2.0.

The great difficulty in parsing v1.1  was because of the complexity of 
the STC schema, which is included. Building a full library requires a 
get and set method for all of these elements, explicit and implicit.

There is a HUGE reduction in complexity of V2.0 because it no longer 
includes the full STC schema, but rather just the elements it needs. 
This is the so-called "veneer" approach -- it looks like STC, but is not 
the whole thing.

However, v2.0 adds in the simpleTimeSeries. So the right comparison is 
here v1.1+STC vs v2.0+sTS
(each sum is complexTypes + simpleTypes + attributes)

VOEvent-v1.1.xsd        12 + 29 + 18 = 59
stc-v1.20.xsd           45 +130 + 34 = 209
Total complexity = 268, with 80% of that from STC


VOEvent-v2.0.xsd        31 + 59 + 31 = 121
simpletimeseries.xsd    31 + 47 + 44 = 122
Total complexity = 243, with 50% of that from sTS

This we see a slight reduction in schema complexity from v1.1 to v2.0.

Cheers
Roy