On Fri, 08 Nov 2002 15:06:57 +0000, Simon Oliver wrote:

> Jacqui Caren wrote:
> > On Fri, 08 Nov 2002 09:28:43 +0000, Simon Oliver wrote:
> > 
> > This is only true if RaiseError(s) is enabled and an exception was raised.
> > 
> True.

My apologies for the delay in responding.

> > It is not true for an subselect inserts or updates that can operate
> > quite sucessfully on empty sets.
> I see your point.
> 
> Is this an ADO only problem - what happens iof you use DBD::ODBC?

We have tested almost 10 ODBC drivers and each one has major faults
even the commercial ones just blow up when used under our DBI portability
test harness.
..
As the target system is production/performance oriented there is a
need for stability and reliability (and reasonable performance).

FYI: I contacted a few of the authors/vendors. The only ones who
did not reply or failed to provide any help were the commercial
suppliers. people such as Tlowery (mr DBD::ADO) and the rest of
the driver writers consitently provide a level of support that far
exceeds the dire verbiage I have seen form the commercial (ODBC) vendors
"support" teams.

In one case I had the salesman emailing me and calling me almost daily
and I had to explain that his product is (currently) useless as it
blows up fatally if you do something as basic as try and bind an undef
to a placeholder. After two weeks with no response (even a "we have your
problem and are dealing with it") for what should be a very basic fault
and remedy (it is usually due to uninitialised buffer with value or length
fetch) I explained that I had ran out time and they had to be excluded from
the shortlist. Even after this telling message I kep getting calls from
the salesman asking when the customer wanted to make a purchase :-)

> Is the problem limitted ot DBD::ADO or is it an ADO problem - does it go 
> away if you use Win32::OLE and create an ADO connection object, etc?

Thomas Lowrey is making/has made changes and we are evaluating these.

There are pros and cons to the change and we are investigating these.

At the mo the choice seems to be one $sth per $dbh and a valid $rv from
execute()s or many $sth's per $dbh and 0 as a rv from execute().

Note that some people have replied stating that they recieve -1 instead
or 0 from execute()s.


Being an ornery sod I want both :-)

Thanks for your reply,

Jacqui


Jacqui Caren, Ingram Group Ltd. [EMAIL PROTECTED]
ph: +44 (0) 1483 8628xx main=00 fax=01 ddi=65
http://www.ig.co.uk/ http://www.sitedirector.org/
http://www.perl.co.uk/


Reply via email to