V4MD - Error -39

[Roland Hildebrandt]

> On 9/7/04 5:36 PM, "Roland Hildebrandt" <r.hildebrandt at sbt.fr> wrote:
>
> > Hi Ruslan,
> >
> > You mean our Database before error -39 occurs is corrupted ? Is there a
way
> > to find out corruption with a crypted database ?
>
> In general yes, Diagnose()
> Just it not correctly report about boolean fields

Yes I do the diagnose on the database but I get messages LIKE "Table has 3
physical records but field has 256" on most of the fields not only boolean.
But I remember some messages on the list where it was said this is normal
with crypted database.
You can see the diagnose at the end of my last mail.

> > Thank's for the advice of XML dump but my problem is not to save some
data
> > in the files.
>
> > I need to understand why we become this error on all accounts databases
> > after 6 months of work.
>
> This could be sys/app crash
>

I don't think so as far as we get the same problem on 20 computers in one
week.
And I can redo the bug again and again using the way I explained in the last
mail.
So the problem seems to be linked to the state of the DB.
The problem is to understand why the DB goes in this bad state after 6
months of use.


> > In fact we have to make the same kind of project for october. So I have
a
> > big problem because I know I will lose all data after six months on one
of
> > the database but I don't know why and I have no workaround.
>
> > That's why it's important for me to understand why this happens.
>
> You should make sure that you have as many flushes as possible.
> This reduce chance of corruption.

Accounts is the only DB we change during execution. We use flush command on
this DB in each script (not each function, but at the end of the script)
which updates it.
This raise me another question : Our other DBs are read only but i have been
surprised that the files (vdb, ind, dat, blb) changes even if not updated. I
have seen this using CVS : I open the DB make a read only request on it then
close DB. After this the file is 'modified' in CVS. What's reason for this ?
Is it linked with some indexes or optimisation for following request ?



Best Regards



Roland Hildebrandt