Breaking out of RecordLock...

[Ruslan Zasukhin]

on 2/18/04 4:47 AM, Ken Ray at kray at sonsothunder.com wrote:

Hi Ken,

> As an example of how we're writing things, in general we use
> SQLSelectRecords() to get data to display in our interface (for
> single-record views), and SQLSelect() for updating/adding/deleting
> records, or for progressive disclosure in list views.
> 
> This brings up a question:
> 
> If I execute a SQLSelect() to get a cursor,

With ReadWrite flag, yes

> then a Cursor_SetField() to set the contents of a field,

> then a Cursor_UpdateRecord() to update the database, but then forget to do a
> Cursor_Remove(), the field in the table has still been updated, right?

Yes, especially after flush()

> I mean *if* there was a way to forcibly remove a locked cursor, and I forced
> this cursor to be removed, the data I'd posted to the table would still be
> there, right? 

Yes.

I think for Vserver itself we can add next feature:
    
    administrator's command "Disconnect All"

On this command server must disconnect all,
And therefore force destroying of each cursor.
For user this will looks as
    disconnect on server failure
    or network failure
    or timeout

I think we cannot preserve connections and destroy just cursors.
This is not correct.


ANOTHER FEATURE which can help diagnose problem:
    ask each opened database how many cursors it have now.
    we can extend DatabaseInfo class.

Igor, please consider this 2 features, and if you think you can do this,
please do this today


-- 
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
-------------------------------------------------------------