Re: scanning user space

[Scott Klement <klemscot@xxxxxxxxxxxx>]

On Mon, 23 Jun 2003, Richard B Baird wrote:
>
> But if you don't think i'm not going to gain anything by loading the whole
> thing at once to a userspace and scanning it from there, then i'll out bail
> on this one.
>

I don't think you'll gain anything by putting it in a user space, but I
don't know for sure.   I've certainly never tried it!  I'd be interested
to see how it turns out.

It seems likely to me that what you'd be doing, when you copy 2million
records from a stream file to a user space, is just copying megabytes of
data (which may not even fit, since user spaces are limited to 16 mb!) and
then scanning it in the same manner that you would otherwise be scanning it....

I guess it seems more efficient to me to deal with it in smaller pieces!

But, on the other hand, that could be the Unix programmer in me talking...
OS/400 can be a strange animal.   I know that the system is better at
dealing with user spaces than stream files (despite the obvious logic that
they're basically the same thing) and it's likely that being able to
directly scan the user space would save you a bit of time in parsing.
The question is, would it save you as much time as you're sacrificing by
copying over 2 million records that you wouldn't otherwise have to copy?

It's an interesting question, yes?