On Friday, March 23, 2012 10:21:07 AM UTC-4, Richard wrote: > > Why are you doing that? > > Richard >
A legacy system, which has stand-alone appservers+databases in geographically distributed locations, with intermittent and low bandwidth connections (think mobile GPRS modems for the kind and quality). I'm building a management system that keeps a complete view of the system. This is done by: 1) a web2py app installed on the legacy system maintains an up-to-date sqlite copy of the existing legacy (postgres) database. 2) this database (.sqlite and .table definitions) is copied to the main management system when bandwidth and connectivity allows it, and placed in its own directory. 3) management system reads all independent databases, consolidates them, computes some data, and makes it available (.sqlite and .table) for standalone appservers to download when they can 4) standalone web2py appservers download updated version when they get a chance, open it and use the aspects of the consolidated data that they need. *THIS IS NOT MY USE CASE*, but an essentially equivalent example is "ice cream truck network". Each truck runs its own inventory. Center keeps an updated view of all inventories, summarizes them, and sends back to trucks. So if you come to a truck on 20th street, and they are out of your favorite ice cream flavor, they can point you to the nearest truck that (as of the last update) had that flavour; All information is available on an as-available-and-possible-to-transfer basis. The original version of my code exported to csv, sent with http, and imported at the other end. The new code is much simpler, more efficient and more robust -- but brought up this snag.
