>Hi Linda,
>
>>One thing I hadn't thought of was the memory that's requird to buffer tables. If I open tables READ ONLY, would that involve some sort of memory saving?
>
>Only do that if you like to have troubles. *g* The readonly flag is one of the settings that are shared across all datasessions, just like the index definitions.

I'm not sure if I'm quite following on this one. Index definition is stored on disk, and you can have the same table open with different orders (and even different non-structural indexes) in different datasessions, or even in the same datasession by means of use...again. Opening a table in use...noupdate should be quite different from having the readonly attribute on the .dbf file, right? It's quite possible to have the same table open once as read/write and once as readonly (ie. with NoUpdate clause).

Since we're talking memory usage here, the only difference would be that the nonupdatable table would not (have to) use buffering. I would speed things up, though, because the OS would deal with such a table more efficiently, not having to apply any locking. I've seen some great speed improvements on older systems (talking 286s and 386s here) when tables were open with NoUpdate clause, but I haven't tested it thoroughly for a number of years, and I wonder if this still holds.