Forum Discussion
Error 11 when writing results to a database.
I am trying to insert results into a sqlite database using the following:
SQLexecute("
INSERT INTO res_rm_markets_oper
(case_id, market, strat_node, oper_scenario, oper_period, bid_up, bid_down)
VALUES (" + caseID + ",?,?,?,?,?,?)",
[array(rm in RM_MARKETS, sn in ST_N, osc in SN_OP_SCENS(sn), op in O_PERS(osc)) rm_bid_up(rm,sn,osc,op).sol,
array(rm in RM_MARKETS, sn in ST_N, osc in SN_OP_SCENS(sn), op in O_PERS(osc)) rm_bid_down(rm,sn,osc,op).sol])This fails with the following error message:
ODBC: Inconsistent array list.
Mosel: Error located at line 452 of `XXX.mos'.
ERROR input parsing returned error code 11! I don't understand what "inconsistent" means here. Both `rm_bid_up` and `rm_bid_down` are dynamic arrays defined on the same sets.
Moreover, the command fails even if I use the same variable on both lines, so the arrays are equal.
It works with only one variable (and one question mark replaced by a number) - both for the first and the second variable.
I have also tried filtering the array using `exists()`, but it does not help.
The weird thing is that we have many such commands in the model from before and I do not see how this one differs from the others.
Any idea what is going on, or how can I debug this?
2 Replies
The indexing sets of the two arrays that are used in the SQL statement are not entirely the same: when processing the following definition:
array(rm in RM_MARKETS, sn in ST_N, osc in SN_OP_SCENS(sn), op in O_PERS(osc))
the compiler will generate new sets for each of the 2 last indices, resulting in something like:
array(RM_MARKETS, ST_N, set of integer, set of integer)
such that they are not identical for the two arrays (even if in the end they have the same content).
A way of avoiding the issue could be to use globally defined sets for these indices and rely on a condition to select the requested values:
array(rm in RM_MARKETS, sn in ST_N, osc in I, op in J | i in SN_OP_SCENS(sn) and op in O_PERS(osc))
with I and J being the actual indexing sets of the target arrays (rm_bid_up and rm_bid_down). If they are not explicitly defined elsewhere in your model, they could be calculated as
I:= union(sn in ST_N) SN_OP_SCENS(sn)
J:= union(osc in I) O_PERS(osc)
Dear Susanne,
thanks for the answer, your suggestion works. :-)
(And looking at the existing code in the model, all DB-writing code uses global sets, so I must have known this at some point, but did not write it down...)
