Fabio,

I don't think that we know enough about VFP's internal memory handling to be able to draw any conclusions from your tests here.

If I was VFP I wouldn't worry too much about memory usage until I got into a crunch, then I would "address" it.
I might also have some timer-based clean-up activity or some low-activity-based clean-up.

We don't even know if VFP 'frees' buffers of tables that have ben closed.
Not to mention what the OS is actually doing with caching of tables! I do know that table data will stay in OS cache even when they are closed in VFP and VFP has done a SYS(1104).

cheers

>Hi Sergey,
>
>yes
>
>but
>
>CREATE CURSOR tab (a i)
>INSERT INTO tab VALUES ( 1)
>INSERT INTO tab VALUES ( 1)
>INSERT INTO tab VALUES ( 1)
>INSERT INTO tab VALUES ( 1)
>INSERT INTO tab VALUES ( 2)
>INSERT INTO tab VALUES ( 2)
>INSERT INTO tab VALUES ( 2)
>
>INSERT INTO tab VALUES ( 3)
>INSERT INTO tab VALUES ( 3)
>INSERT INTO tab VALUES ( 4)
>
> then these A
>
>INDEX on a TAG i1
>FOR ju=1 TO 100000
>INDEX on a TAG i2 && del tag, bloat memory and bloat cdx
> ( don't reuse cache !! )
>next
>* have more 100MB memory used ( table have 10 rows !!! )
>DELETE TAG i2 && del tag and free memory
>DELETE TAG i1 && close cdx
>
>and these B
>
>INDEX on a TAG i1
>FOR ju=1 TO 100000
>SYS(1104) && free memory ( garbage collection )
>INDEX on a TAG i2 && del tag, bloat memory and bloat cdx
>next
>DELETE TAG i2
>DELETE TAG i1
>
>and these C
>INDEX on a TAG i1
>FOR ju=1 TO 100000
>INDEX on a TAG i2 && bloat memory and bloat cdx
>DELETE TAG i2 && del tag and free memory
>next
>DELETE TAG i1
>
>are equal.
>
>yes for cdx size, but not for VFP Memory Allocation ( use WTM ).
>
>then exist 3° CDX Bloat problem ( code A ):
>
>If redefine tag index then you have:
>
>cdx Bloat ( little problem with 100GB HD today)
>memory Bloat ( big problem )
>
>I suspect this Memory Bloat exist on other situations.
>
>Need one global timer with periodical SYS(1104) command ?
>
>Fabio