Vlad Horsun wrote:

>>Personally, and reasonable people can (and will) differ, that temporary
>>tables aren't worth the tradeoff necessary to implement them.
>>
>>
>
> Which tradeoff do you (and "reasonable" people) mean ?
>
>
>

Your implementation replaces a page number, a 32 long, with a structure
containing a segment and page number. Every place a page number is
referenced, hundred of places, maybe more, is going to have to be
changed to initialize the structure. And every time the structure is
passed, it's going to require more instructions in the calling
sequence. There are other possible implementations with a lower cost,
but the one you suggest rubs up against a subjective judgment: Is the
runtime cost of a feature that everyone pays all the time fairly
balanced the the benefit of the feature? My take, and it's just my
personal opinion, that in this case, benefits don't justify the cost of
that implementation. This isn't to say that I'm against global
temporary tables. They're a useful feature, and as you point out, part
of the standard. I just wish we could figure out an implementation that
didn't have such a high engineering, maintenance, and runtime cost.

--

Jim Starkey
Netfrastructure, Inc.
978 526-1376