corrupted databases - we need a tool!

[Robert Brenstein]

>On 10/5/04 4:47 PM, "Robert Brenstein" <rjb at rz.uni-potsdam.de> wrote:
>
>>  While true repair might not always be possible, such a tool could
>>  offer a rebuild function, which creates a new copy the db by
>>  extracting all that is possible to extract from the old one, thus
>>  rescuing as much as feasible, hopfully reaching deeper than xml dump
>>  can do. Ideally, these things should be part of kernel self, not a
>>  standalone tool, so we can use them directly in all environments,
>>  allowing us to include recovery options in our products.
>
>Aha, I see idea, Robert!
>
>Yes this is good idea.
>Really good.
>
>It is possible to TRY recovery any data which are readable,
>And code which will read this should have many protection from failure.
>
>
>>  Trasactions are still distant future while a repair tool has been
>>  talked about since early 2002 at least (posts dealing with db
>>  corruption go back to 1999 upon quick search) and was expected for
>>  1.x technology to follow on the heels of the diagnose coming to life.
>
>
>>  But diagnose self still has a few glitches and is not available as
>>  part of kernel functionality. As long as Valentina does not clean up
>>  and build up certain aspects that have been asked for over the years,
>>  it can't shake its shareware-only image and step into the ring of big
>>  boys.
>
>Right.
>
>But as I have told, we have developer which have many years work with
>Sybase. He say that IF Sybase db corrupts, it also is not many chances to
>recover it.
>
>This is complex task, and it cannot be solved in 5 minutes.

Sure. But last more extensive discussion was in early 2002 and a few 
releases came out since then. DB integrity check was first delayed 
because it required file format change. After that happened, it kept 
waiting for diagnose to grow up. So, a couple years later, 1.x 
reached its life end so do speak and still can't check integrity when 
opening dbs and recovery is restricted to xml dump.

Let me be clear that I/we have no problem accepting the fact that 
such a tool can't always be successful. It will be like using Norton 
Disk Doctor for diagnosing and repairing hard drives. Norton can 
correct few things in place, others have to be recovered to another 
drives, but sometimes it can't do anything useful. But at least it 
lets us try.

The fact that Sybase can't do it well is not an excuse. I haven't 
used Sybase but spent many years with FileMaker and never had any 
serious corruption issues. And the few times I crashed, the built-in 
repair of FileMaker fixed the db up.

I know that you are more of a visionary and concept-man and thus 
things like recovery, documentation, examples, etc tedious aspects 
are not exciting for you (and waste of your talents as well), but now 
that you have more people on your team, you can set someone up to it.

Robert