[Gnumed-devel] re xml-rpc

[sjtan]

I worked with omni corba ( which has a lgnu license) and xml-rpc test
code , and found the omni idl generated types richer and more satisfying
to program. omni corba seemed faster too. The disadvantage is that
omni corba is yet  another separate package from the python 
distribution; if using a corbamed specified interface a) it may not be 
suited ambulatory care needs b) it may be claimably owned by the 
specifying parties, even though they explicitly state they will not 
claim on implementations c) its Observation COAS  and PIDS 
specifications are not too difficult to implement, but no substantial 
Lexicon Query Service seems to exist freely and publically available, 
which seems to suggest that all the
hard programming work is in terminology classification , ontology and 
decision support, and no one wants to do it for free (Protege is "free", but
it's in evil java)
Corba also isn't built upon http or https , so clients need heavy corba 
libraries (albeit, they are more type checked, and calls are faster).
Other than being built-in to python as the only available 
remote-procedure call library , I'm not sure what the advantages of
xml-rpc are.  Is Gnome an example of CORBA's eventual non-robustness?
Having played around with other "free for study, not for use" such as
java and qt , and found the licensing paranoia immovable (but hey, RMS 
succeeded with this attitude) ,  why not study something like qt's 
implementation of  data-aware widgets ( which is a collaboration of
gui and non-gui  sql classes) , and build a data-aware wxWindow widget
collaboration ? That might mean leaving things inside cursors for 
viewing, but doesn't stop validation/triggering from cursors with 
business objects. That would fit in with the roll-your-own except for 
gmx-datetime(.com)/PyPgSQL/PostgreSQL/wxWindows   packages  regimen, 
wouldn't it?