Re: RSE performance against very large source files

[Joe Pluta <joepluta@xxxxxxxxxxxxxxxxx>]

Dave Shaw wrote:
> This is one of the issues that's killing acceptance of the product in this shop.  It would be very hard to justify NOT having a copy of PDM for everyone with this kind of issue, and if the company has to buy everyone a copy of ADTS, why should they also buy RDi?  Yes, I know why, but it hasn't been an easy sell getting people to use WDSCi for the price of a decent PC and some learning time; now I have to justify buying it, too.
>   
Have you raised this with IBM, either as a PMR or a DCR?  You seem to 
have a legitimate issue, and it's an underlying system issue.  Two 
things are immediately apparent, at least to me. First, there is 
something ridiculously wrong if it takes 39 minutes to do a DSPFD on a 
file with a few thousand members (or even 20,000 members).

In my unofficial test I just ran, it took less than 15 minutes to add 
15000 members to a file.  Next, I did a DSPFD on it, and the list came 
up in about 30 seconds. I then went into WDSC and open a connection to 
my machine.  I added the library to my library list, and then expanded 
the library.  Two seconds.  I then expanded the source file.  12 
seconds.  Now granted, these were empty members.  There may be other 
issues that come into play when there are lots of records in the source 
(DSPFD definitely displays number of records and deleted records; if it 
calculates those dynamically at display time, that could explain some of 
the overhead).

Even so, 39 minutes to display a member list seems excessive.  Is there 
something else at play here?  An exit program, perhaps?  Some sort of 
bizarre auditing issues?  Have you asked IBM about it?

But there's a second issue, which is probably more to your specific 
point.  Evidently PDM manages to do something under the covers that 
allows it to get at a list of members much more quickly than DSPFD.  If 
that's indeed the case, then maybe there's a call for some sort of 
"paged widget" in RDi that can directly access the same functions that 
PDM uses.  There would be limits, such as not being able to filter 
without a performance hit (because as you know, filtering in PDM takes a 
healthy chunk of time as well).  But if enough people were to tell 
George Farr that this is a deal-breaker, then he might be willing to 
devote some resources to it.

Joe