Substr/Instr/analytics question

Question

I have a table like this on 9i (soon to be 10g)
CREATE TABLE BASELINE_TESTRUN
RUNID VARCHAR2(42 BYTE) NOT NULL
with data like this:
insert into BASELINE_TESTRUN values ('DEV1-XXX-01');
insert into BASELINE_TESTRUN values ('DEV1-XXX-03');
insert into BASELINE_TESTRUN values ('DEV2-XXX-01');
insert into BASELINE_TESTRUN values ('DEV2-XXX-02');
insert into BASELINE_TESTRUN values ('DEV2-XXX-03');
insert into BASELINE_TESTRUN values ('DEV2-XXX-05');
insert into BASELINE_TESTRUN values ('DEV2-XXX-06');
insert into BASELINE_TESTRUN values ('DEV2-XXX-06');
The output should be the next higher number of the last substring in runid
DEV1-XXX 04
DEV2-XXX 07
There may be missing no, and/or duplicates.
This was my Query for this - It works but seems rather clumsy. I would like to have done it in 1 SQL (avoiding the select from (select)):
select runid , max (next_runno) next_runno
from
select substr(runid,1,instr(runid,'-',1,2)-1) runid,
max(substr(runid,instr(runid,'-',1,2)+1))
over(partition by substr(runid,1,instr(runid,'-',1,2)-1)) + 1 next_runno
from baseline_testrun
where instr(runid,'-',1,2) > 0
---where runid like 'whatever%'
group by runid;
Alternate solutions would be appreciated ... and NO I did not define that table like this.
I would have preferred a sequence and a keys that was not composit (Codd/Date, pls. forgive the creators of the table)
best regards
Mette

Answer 1

You are grouping you data. In this case an aggregate function could be more helpful than an analytic function.
In 9i already you are be able to use the KEEP syntax for aggregates.
Look at this example
SQL> select substr(runid,1,instr(runid,'-',1,2)-1) group_run,
  2  to_number(
  3  max(substr(runid, instr(runid,'-',1,2)+1))
  4  keep (dense_rank last order by substr(runid, instr(runid,'-',1,2)))
  5  )+1 max_run
  6  from BASELINE_TESTRUN
  7  group by substr(runid,1,instr(runid,'-',1,2)-1);
GROUP_RUN                                     MAX_RUN
DEV1-XXX                                            4
DEV2-XXX                                            7
SQL> Dimas Version is better, since it avoids the keep syntax, that is not really needed for this special case.
Message was edited by:
Sven W.