Re: Retrying a fetch after an error, without restarting the whole loop?

On 11/9/06, Bart Lateur <bart.lateur@xxxxxxxxxx> wrote:

On Wed, 8 Nov 2006 23:26:02 -0800, Jonathan Leffler wrote:
> Bart Lateur <bart.lateur@xxxxxxxxxx> asked:
>>And 2), in a fetch loop, is it possible to adjust a property like
>> {ReadLongLen}, and retry the same fetch without restarting the whole
>> loop? Because this error typically happened several minutes into the
>> loop.
>
>Highly unlikely. The data has been fetched - and truncated. There's not
>usually a way to refetch the same row - unless you have a scroll cursor,
and
>DBI doesn't have support for those.

I can see that. Well I'm thinking of the following solution next:
retrieve extra data to identify the row that went wrong and collect
them, keep going on with the rest of the records, and individually fetch
the previously failed ones afterwards.

After a failure, I can go on with the next records, can't I? And
changing ReadLongLen, is that acceptable for the remainder of the loop?



That works - I've used it in (non-Perl) code for aeons (dating back to
1986). The primary demerits are (1) two fetches for each row, and (2)
establishing the unique identifier column or columns for the data.

And yes, I can think of no reason why a driver that honours ReadLongLen
would not adapt to changed values as the loop continues. Since
DBD::Informix doesn't pay attention to ReadLongLen in the first place, it
isn't much use looking at that code - and other driver writers would have to
answer for their code.

--
Jonathan Leffler <jonathan.leffler@xxxxxxxxx> #include <disclaimer.h>
Guardian of DBD::Informix - v2005.02 - http://dbi.perl.org
"I don't suffer from insanity - I enjoy every minute of it."

Relevant Pages

  • Re: Self Joins and optimization
    ... What I'm interested in exploring is your assertion of the ... non-existence of a set oriented solution to the problem, ... while x.ArrivalTime is null and x%FOUND loop ...
    (comp.databases.theory)
  • Re: Bulk Collect without LIMIT
    ... Oracle, with 10g, merged the BULK COLLECT and FETCH to use the same mechanism. ... If you just say OPEN CURSOR ... ... for i in 1..500 loop ...
    (comp.databases.oracle.misc)
  • Re: Ada DB bindings and APQ
    ... > supported on sequential connections, because it is not known if all ... > tuples have been fetched until you try to fetch the next one past the ... > condition for a loop. ... Fetch would raise an exception (it cannot be ...
    (comp.lang.ada)
  • Re: -1555-ORA-01555: snapshot too old: rollback segment number 7 with name "_SYSSMU7$" t
    ... it happen to fetch something from a big result set in a loop and commit ... Or does it happen that the query runs longer than ... Well, if this script receives ORA-1555 on regular basis, then you ...
    (comp.databases.oracle.server)
  • Re: Calling a SP inside a cursor loop..
    ... every loop iteration of the cursor. ... Fetch next From EffectiveDate_Cursor Into @FLD1,@FLD2 ... and If the Fetch stmt is below the begin Stmt, the loop iterations are ... the proper way to program a cursor loop is: ...
    (comp.databases.ms-sqlserver)