FWD » Core Development » Database

• methods
  • FIRST-OF (none/none)
  • LAST-OF (none/none)
  • REPOSITION-BACKWARDS (full/basic)
  • REPOSITION-FORWARDS (full/basic)
• attributes
  • CACHE (maybe, very implementation-specific) (also noted in #6459)

Note that FIRST-OF/LAST-OF are already implemented; the gap marking is wrong. It is strange, because these are gap marked in the expression.rules attrs map, even though they are methods which accept a parameter. When reports are run, instances appear as METH_LOGICAL, but pick up the the gap marking from the attrs map (none/none at the time of this writing for both) from the expressions.rules gap marking rule set.

Alexandru, do you have any insight about the reposition methods in the first note? I recall seeing mentions about reposition, in some tasks (where also Stanislav was working).

REPOSITION-BACKWARDS (full/basic)
REPOSITION-FORWARDS (full/basic)

AFAIK, these are both supported at conversion time and run-time. I made unit tests on #9724 for both these constructs on single-table queries (only scrolling, indexed-repositioned/non-indexed-repositioned, browsed/non-browsed). All worked OK. I was planning to extend the work to multi-table queries, but I doubt there is a lack of functionality there.

The only concern may be related to performance, because these constructs do next, next, next or prev, prev, prev to succeed. But I can't find a proper alternative ATM.

PS: I think the constructs are :REPOSITION-BACKWARD and :REPOSITION-FORWARD.