Ruslan,
no, I'm not using a cursor to import, I'm doing it with
BaseObject.AddRecord(). Strange thing is, (I guess) the indexes should
exist all the time, since I'm "re-using" the same database for several
months now, always doing the same at the end of the month:
bo.DeleteAllRecords()
for n=1 to xxx
bo.Field("blah").SetString(myBlahBlahValue)
... blah blah ...
bo.AddRecord
next
One of the indexed fields is defined as unique, though... I have the
feeling that maybe when deleteing all records, some of the index data is
remeining intact, and Valentina has to search through it to check if the
value is truly unique... ???
Also, I have noticed that I can improve the overall importing speed if I
call Flush() from time to time, say every 1000 records or so.
BTW, when writing "importing", I don't mean Valentina's own Import
method, but rather AddRecord().
Cheers,
Frank+++
Ruslan Zasukhin wrote:
>
> on 10/31/01 14:28, Frank Bitterlich at bitterlich .AT. gsco .D.OT de wrote:
...
> > I have noticed that when I do this the "normal" way (dlete all records,
> > flush, start importing) the importing becomes very slow after approx.
> > 200.000 records (20 rps and less instead of 350 rps initially). But when
> > I do this, but abort the import shortly after beginning, and start it
> > over (again delete all records, flush, start importing), the
> > records-per-second rate doesn't drop that substantially; at 500.000
> > records, I still get about 200-250 rps.
...
> indexes....it looks like in first case indexes exists and Valentina add
> record and index it immediately.
>
> after you deleteAll records you use the same cursor ?
>
> you import via cursor ?
--
Günter Schmidt & Co. oHG
Frank Bitterlich eMail: bitterlich@gsco.de
Schlosserstr. 2-4 WWW: http://www.gsco.de/gsco
D-60322 Frankfurt Tel.: 069 / 156809-29
GERMANY Fax: 069 / 156809-28
©2001 Frank Bitterlich |
