Page 1 of 1
Browse slow change of active cell
Posted: Thu Nov 04, 2010 11:27 am
by solvox
Hi all,
Open sample samples/browse/demo_5.prg
change line 21: Use demo5 -> Use demo5 shared
Start program, make 1 million records, click on some cell.
On my system (win7) it's very slow on changing from active cell to new one.
Anyone else have that behaviour?
Thanks
Alen
Re: Browse slow change of active cell
Posted: Thu Dec 09, 2010 4:55 pm
by nguyenchiduc
solvox wrote:Hi all,
Open sample samples/browse/demo_5.prg
change line 21: Use demo5 -> Use demo5 shared
Start program, make 1 million records, click on some cell.
On my system (win7) it's very slow on changing from active cell to new one.
Anyone else have that behaviour?
Thanks
Alen
You are right
I test the file demo5.dbf, with 2 million records and 381M.
Use demo5 / *
It is good * /
Use demo5 shared / *
It is slow * /
Reason is the command line 22:
Set Index To demo5
And when I:
Use shared demo5
/ *
Set Index To demo5 * /
It as fast as
Use demo5
Re: Browse slow change of active cell
Posted: Fri Dec 10, 2010 2:46 pm
by nguyenchiduc
I tested with
10 million records and it works but with
100 million it doesn't works, very very slow.
BROWSE command is working fine in VFP
Note :
I do :
Use demo5 or
Use demo5 shared
I have not test :
Re: Browse slow change of active cell
Posted: Mon Dec 13, 2010 10:27 pm
by mrduck
I confirm that adding SHARED makes the demo almost unusable...
but if you use DBFCDX it is usable again... but then you will notice that the code underlying the BROWSE is a little bugged in record positioning for screen updates... so it is not usable in production...
Any volunteer to do some code checking ?
Re: Browse slow change of active cell
Posted: Mon Dec 13, 2010 11:12 pm
by mrduck
mrduck wrote: but then you will notice that the code underlying the BROWSE is a little bugged in record positioning for screen updates... so it is not usable in production...
Any volunteer to do some code checking ?
After a bit of code cleanup the problem in screen updates disappeared... perhaps I did introduce the bug in some testings... I will check tomorrow...