D
dbeavon
Guest
@ske That is a very good thought, and I have made some attempts in the past to leverage schema caching for these types of things. But schema cache doesn't work well (or at least not consistently well) from my experience. There must be a lot of special-case rules in the AVM about when/where it is allowed to use the schema cache. Even "DYNAMIC QUERIES" will ignore the schema cache at run-time, and that seems to be the scenario where the cache is intended to have its greatest effect. It has been a while but I think I had tried improving the performance of PDSOE compiles by using schema caching and wasn't very successful. I suspect that ABL compile operations don't use the cache in general. Perhaps it is only used by the AVM's runtime validation, whereby comparisons are made between the r-code and the database structure. The docs say that the cache is a "complete schema or subschema for a single OpenEdge database", but I don't know if it is used as the "authoritative" schema, or as a perfect substitute for accessing the database directly.
Continue reading...
Continue reading...