Hi is there anything similair to the sql distinct directive in DBisam
tables.
I dont want to use sql but I would like to only show one instance of
a name like in a names table I only need one 'Young' even if there are
20 or more. I cant make it a unique index because of data loss.
Rita

Rita


You could probably cobble something together using the OnFilter event, or using a mixture of sql and tables but ASAIK there's nothing you can use directly.

Roy Lambert [Team Elevate]

Thanks Roy I found this and will check it out 1st lazy bitch
http://www.grebarsys.com/

SFD sounds good but is it I will soon find out.
Rita

"Roy Lambert" <roy.lambert@skynet.co.uk> wrote in message
news:756FE26C-F65C-4C3A-AFAE-AE1DF1A75211@news.elevatesoft.com...

I love it when something goes right I dl'd the demo was pleasently suprised
to
see a DBisam demo but thats not all it was an IW demo toboot.
My lucky day its also half price yabbadabbadoo.

Rita

Rita


If all you wanted were the distinct values you'd probably be better off using SQL:

SELECT DISTINCT Surname FROM Table

That way you're assured of compatibility as DBISAM gets upgraded or you switch to a different database (eg ElevateDB)

Roy Lambert [Team Elevate]

"Roy Lambert" <roy.lambert@skynet.co.uk> wrote in message
news:4B1D4392-2C5F-468F-9E59-E727BAAE1A93@news.elevatesoft.com...
>
> If all you wanted were the distinct values you'd probably be better off
> using SQL:
>

Sorry Roy you have gone full circle on me there.
Bad SQL naughty SQL

> SELECT DISTINCT Surname FROM Table
>

I know how the SQL distinct works I want to avoid it as
a 1gb of records is slowwwwwwwww

SuperFastDistinct seemed a good idea see the elevate.public.thirdparty
dated 25/04/2006 now I'am not sure.

"A quote by GregF
I don't understand why I would want to spend so much money on a product
that a simple and coding solution can achieve, I mean geez in effect
all you are doing is positioning on the first key and then seeking the next
value
and the fact it needs hierarchical indexes means the product is of no use to
me unless I actually have an index on the fields++ required"

Barry the guy behind it has not even updated the site it shows a discount
from 2006 worded as June 06 sorta suckered me so no dough for him.
All I have to do now is figure ouy how to do it from what GregF says.
Rita

Rita


DBISAM 4.25 on a Dell Vostro 1700 (2.5Ghz duo)

select distinct _whofrom from mandn

4.5 secs returning 11867 records from 262500 records (.dat 114Mg, .blb 4Gb .idx 253Mb)

Roy Lambert

"Roy Lambert" <roy.lambert@skynet.co.uk> wrote in message
news:54B62C8F-C3A6-44A6-BFE2-F3214D384219@news.elevatesoft.com...
> select distinct _whofrom from mandn
>
> 4.5 secs returning 11867 records from 262500 records (.dat 114Mg, .blb 4Gb
> .idx 253Mb)
>
So slowwwwwwwwwwwwwwww
I guess i cant read 11867 records so why worry about the 4.4 secs extra.
CS read is not a problem but what about writes and 50+ users its ok
bench testing one but in a real data mine the miners hate SQL lags for
what should be simple code.

Rita,

Index the table by the key you are looking for. Start at the first
record & get your first distinct value. Do what you want with it. Change
the last character in the distinct value to the next ordinal value. For
example, 'Jones' would become 'Jonet'. Do a find nearest on this value
to get the next distinct key value. Repeat until no more are found.
Probably not necessary for text data, but you could be safe and check
for a string ending in a character 127 or 255 and in that case append a
CHR(1) instead of changing the last character.

Sean


Rita wrote:
> "Roy Lambert" <roy.lambert@skynet.co.uk> wrote in message
> news:4B1D4392-2C5F-468F-9E59-E727BAAE1A93@news.elevatesoft.com...
>> If all you wanted were the distinct values you'd probably be better off
>> using SQL:
>>
>
> Sorry Roy you have gone full circle on me there.
> Bad SQL naughty SQL
>
>> SELECT DISTINCT Surname FROM Table
>>
>
> I know how the SQL distinct works I want to avoid it as
> a 1gb of records is slowwwwwwwww
>
> SuperFastDistinct seemed a good idea see the elevate.public.thirdparty
> dated 25/04/2006 now I'am not sure.
>
> "A quote by GregF
> I don't understand why I would want to spend so much money on a product
> that a simple and coding solution can achieve, I mean geez in effect
> all you are doing is positioning on the first key and then seeking the next
> value
> and the fact it needs hierarchical indexes means the product is of no use to
> me unless I actually have an index on the fields++ required"
>
> Barry the guy behind it has not even updated the site it shows a discount
> from 2006 worded as June 06 sorta suckered me so no dough for him.
> All I have to do now is figure ouy how to do it from what GregF says.
> Rita
>
>
>
>

Rita


OK so what is the actual scenario. You want reads & writes over the internet? What size database in records, how many unique records do you expect to find etc?

Roy Lambert