cursor/api

Ralf Sander ralf at end-if.de
Thu Feb 10 02:15:58 CST 2005


Am 10.02.2005 um 01:32 schrieb Ruslan Zasukhin:

> On 2/10/05 1:42 AM, "Ralf Sander" <ralf at end-if.de> wrote:
>
>> Cursor way:
>> I need to use "knolocks" because "kreadonly" lets only one user read a
>> record.
>
> NO!
>
Ok, I only did a single user test. But here it behaves like this:
If I get one cursor with "kreadonly", I can't get another one.

> Many users can read. Nobody will not be able write.
>
>
>> Works good so far. If a record has been deleted outside the cursor, 
>> the
>> record has just no values in the cursor, the record don't become nil.
>
>> Also no problem with reused RecIDs. Will I get the same results with a
>> served database?
>
> Yes, but you describe strange things.
Do you mean the upper (no values in record) or the lower things?
>
>> But I see a possible problem with speed. I have to store a sql search
>> request as a string for the current view of one user. If the user adds
>> a (existing-not new)  record to the current view, the sql request has
>> to be modified and re-performed. Could this be a problem, if the
>> current search is very complex respectively would the api way be much
>> faster?
>
> If you use Vcursor and add records using Vcurosr.AddREcord()
> Then new record is right here.

Yes, a new record. But I can't add an existing record to the cursor and 
I can' re-sort it.

Cheers,
Ralf



More information about the Valentina-beta mailing list