After talking to Jeff Kantor I slightly rearranged the python lsst
framework code to more closely reflect the EA model. The new
arrangement is:
lsst
apps
common (Common Pipeline Components)
(non-middleware-related application support code)
PySextractor
fw (Framework)
(middleware-related application support code)
Policy.py
Catalog/
Collection/
Image/
mw
pipeexec (pipeline execution; was pipe)
queue.py
I have implemented the changes but not yet tested them.
I believe the python fits collection software should go in
apps.common. This is because there is no corresponding entity in the
Domain model; it is lower level support code to support classes from
the model.
Jeff also suggested I look at the svn layout to improve the match of
the middleware layout to the organization of use cases in the EA
model. The following is my suggested structure.
middleware
das (data access services)
catalog (catalog construction)
query (querying)
ingest (data ingest)
dps (distributed processing services)
control (Control and Management)
pipeconst (Pipeline Construction)
pipeexec (Pipeline Execution)
security
sysadmin (Sys Admin and Opns Services)
ui (User Interface Services)
Also...might we consolidate "Pipeline Construction" and "Pipeline
Execution" use cases in the EA model into one package called
"Pipeline"? That would simplify the structure in the EA model and in
the svn repository.
-- Russell
_______________________________________________
LSST-data mailing list
[email protected]
http://www.lsstmail.org/mailman/listinfo/lsst-data
