[DAS2] biopackages server UML

[Allen Day]

Hi,

I'm attaching my first draft for the UML of a server rewrite.  Aside from
all the spec churn, there are two main types of requests that need to be
handled that spurred me to do this rewrite.  The third reason I'm doing this
is to rework the caching mechanism on the server.  With the current code
base there is a lot of custom table clustering and denormalization to get
decent performance out of the Chado database.  I did some experimenting
(discussed in an earlier thread and on conf. calls) with a "tiling" or
"block" caching strategy of cache that turns out to work really well, and I
wanted to integrate that with the writeback functionality.

1) tighter integration of writeback, including locking.
2) configurability of feature types to be
  * dynamic (e.g. for on-the-fly gene prediction)
  * non-cacheable
  * cacheable
3) caching
  * segment range/type tiled caching
  * ability of writeback events to trigger cache flush events

See attached UML.  There is a .zuml file, you can view/edit with Poseidon,
or if you need a .xml I can send another attachment.

-Allen
-------------- next part --------------
A non-text attachment was scrubbed...
Name: das2_refactor.zuml
Type: application/octet-stream
Size: 34991 bytes
Desc: not available
URL: <http://lists.open-bio.org/pipermail/das2/attachments/20061010/5d7a1c0b/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: das2_refactor.png
Type: image/png
Size: 130013 bytes
Desc: not available
URL: <http://lists.open-bio.org/pipermail/das2/attachments/20061010/5d7a1c0b/attachment.png>