Database Size?

[Ruslan Zasukhin]

On 8/21/04 2:53 PM, "Frank Schima" <frank-list2 at mindstarprods.com> wrote:

>> On 8/21/04 6:53 AM, "Frank Schima" <frank-list2 at mindstarprods.com>
>> wrote:
>> 
>>> Really! This is very interesting. Do you remember that I had a problem
>>> with a database becoming corrupt because the segment size was too
>>> small? I was trying to use 128 *bytes* and I saw the problem. I have
>>> used 512 bytes successfully though. It's too bad because 4KB becomes
>>> the smallest record size then for a single table database.
>>> 
>>> So I guess there is no way to support smaller than 4kB segment size?
>> 
>> You can specify less size.
>> 
>> And this ALSO will reduce the initial size of db.
>> Db with 2KB with be 2 times less than db with 4KB.
>> 
>> I have to think more...my words are above are wrong.
>> 
>> We can have db.segment e.g. 1KB.
>> So on one physical page of OS file will be located 4 Valentina
>> segments.
>> This still will work.
>> 
>> And if we want to read 1 Valentina segment then OS will load into RAM
>> all 4
>> segments. Because they are on single page of OS file.
>> 
>> I don't know why there was problems with so small segment size.
>> Probably because FAT also is built on segments, and if you have 128
>> bytes
>> segment, then one FAT page can keep refs of only 32 data-pages.
>> Probably
>> this cause very fast some problems.
>> 
>> I remember that with size of segment is based some limit on size of db.
>> Hmm, we have never talk about this...because limit this limit is very
>> high
>> even for 1-4KB. I need refresh my memory on this.
> 
> I was looking back at my small segment size project. Here is the weird
> thing. When I set the segment size to 128 bytes and 1 file, then it
> creates a corrupted database. But when I change it to a 4 file
> database, then it was ok! For a single file database, I had to set the
> segment size to 4 kB (or larger) in order to create a database that was
> not corrupt. I had sent you the project and file before, let me know if
> you need them again.

I think that was FAT problems/bugs.


-- 
Best regards,
Ruslan Zasukhin      [ I feel the need...the need for speed ]
-------------------------------------------------------------
e-mail: ruslan at paradigmasoft.com
web: http://www.paradigmasoft.com

To subscribe to the Valentina mail list go to:
http://lists.macserve.net/mailman/listinfo/valentina
-------------------------------------------------------------