[V4RB 0000242]: Database.Open crashes for my NEW database

[Frank Schima]

Hi Ruslan,


On Jan 1, 2005, at 11:22 AM, Ruslan Zasukhin wrote:

> On 12/31/04 11:22 PM, "Ruslan Zasukhin" <sunshine at public.kherson.ua>  
> wrote:
>
>> On 12/31/04 10:31 PM, "vbt at paradigma.ukrcom.kherson.ua"
>> <vbt at paradigma.ukrcom.kherson.ua> wrote:
>>
>>> The following issue has been SUBMITTED.
>>> ===================================================================== 
>>> =
>>> <http://paradigma.ukrcom.kherson.ua/bt/view.php?id=242>
>>> ===================================================================== 
>>> =
>>> Reported By:                Frank Schima
>>> Assigned To:
>>> ===================================================================== 
>>> =
>>> Project:                    V4RB
>>> Issue ID:                   242
>>> Category:                   API
>>> Reproducibility:            always
>>> Severity:                   crash
>>> Priority:                   normal
>>> Status:                     new
>>> ===================================================================== 
>>> =
>>> Date Submitted:             12-31-2004 22:31 EET
>>> Last Modified:              12-31-2004 22:31 EET
>>> ===================================================================== 
>>> =
>>> Summary:                    Database.Open crashes for my NEW database
>>> Description:
>>> Attempting to open my newly created database with ViSQL 2 causes a  
>>> crash.
>>> I'm sending a project that creates the database to Ruslan.
>
> Aha, Frank,
>
> We have study this problem, and we found that problem is that you use  
> small
> db.SegmentSize 2 or 4 KB.
>
> If set it 8 or 16 or 32 then all works.
>
> I will try fix this bug now.

OK, that gives me an easy workaround. Thanks!

The real problem I'm trying to track down is that an UPDATE ... WHERE  
... SQL is causing a crash in that database and I cannot debug the  
problem with ViSQL. I'm having trouble trying to replicate the problem  
though, because it works fine on an empty database (of course nothing  
is changed in that case, but it does not crash!) and it also works when  
I add a few hundred dummy records. Maybe a larger segment size will  
solve this problem too.

I will keep studying this and let you know if I can isolate anything.


Best regards,
Frank Schima