Re: WDSCI-L Digest, Vol 6, Issue 70

[Jon Paris <Jon.Paris@xxxxxxxxxxxxxx>]

On 4-Feb-08, at 12:47 PM, wdsci-l-request@xxxxxxxxxxxx wrote:

> "Evidently PDM manages to do something under the covers that
> allows it to get at a list of members much more quickly than DSPFD."

Well PDM doesn't have a lot of formatting to do and all it needs is a  
list of names.  DSPFD has to "dig deeper".

WDSC/RDi is still a huge disappointment in this regard.  CODE/400  
still outperforms it when it comes to retrieving lists.  I guess part  
of it is that once the data is received it has to be wrapped in XML  
for storage.  Or is the XML built on the host?  In which case it may  
well be Java performance on the host that is the issue.

There was also a problem a while back where IBM discovered the  
algorithm used when searching cache was horribly inefficient when the  
cache became large.  That I believe was fixed - but is this possibly  
related?  In other words as the list gets really large the storage of  
each new item becomes less and less efficient?

Of course I can't contemplate why anyone wants to keep so many source  
files in a single file in the first place but that's just me - and  
this is a problem even on more moderately sized lists.

Jon Paris

www.Partner400.com
www.SystemiDeveloper.com