Subject Re: [IBO] development
Author Geoff Worboys
Geoff Worboys wrote:
> Delphi 2010 / IBO v4.10.alpha

I realised there were a few more things I should explain about
the performance tests I reported in the previous post.

My Delphi 2010 tests were done using my experimental version
of IBObjects (a unicode version). This version of IBO includes
my prepare optimisation but does little that should effect the
transfer of row data. (My previous trials of prepare times
suggest non-optimised versions of IBO may see prepare times of
up to 5-10x those that I report, depending on many factors.)

Note that I did also compile under Delphi 6 using IBO v4.8.7
(although once again this IBO had my optimisations installed).
The times were very similar - it was gratifying to see that
the unicode version had not done nasty things to performance
of such basic operations (nor should it, as the data itself
never gets converted in this transfer processing).


Hans, it occurs to me that if your 128M table had a large
number of small tables then the longer prepare times of the
normal version of IBObjects could indeed blow-out transfer
times... but I would have thought it would have to be an
extreme situation to see the sort of times you are seeing.

--
Geoff Worboys
Telesis Computing