Subject | Re: [IBO] Blob problems (IBO and FB 1.5 RC8 -RC9 possible bug) |
---|---|
Author | mp527 |
Post date | 2004-02-19T04:06:14Z |
Helen,
I made the changes you suggested and it made no difference.
When I use the modified code to store a blob, the fbserver process
keeps rising, I had it up to 50mb with ease.
I really think it's a bug in FB 1.5 RC 9.
I also noticed new behavior with RC9, before (RC8) when I
disconnected from the server the FB server process would release the
memory used to store the blob, but now it does not seem to happen.
I will pull out the code and create a test app with DB and post it
on the dev list.
If this is a bug, it is pretty serious because ISAPI dlls and apache
DSO cache data connections for long periods of time, and if the web
server uses blobs to store text or session info, this could be
serious drain on server resources.
Thanks for your help, your suggestions do make my code more
elegent :-)
I made the changes you suggested and it made no difference.
When I use the modified code to store a blob, the fbserver process
keeps rising, I had it up to 50mb with ease.
I really think it's a bug in FB 1.5 RC 9.
I also noticed new behavior with RC9, before (RC8) when I
disconnected from the server the FB server process would release the
memory used to store the blob, but now it does not seem to happen.
I will pull out the code and create a test app with DB and post it
on the dev list.
If this is a bug, it is pretty serious because ISAPI dlls and apache
DSO cache data connections for long periods of time, and if the web
server uses blobs to store text or session info, this could be
serious drain on server resources.
Thanks for your help, your suggestions do make my code more
elegent :-)
--- In IBObjects@yahoogroups.com, Helen Borrie <helebor@t...> wrote:
> At 05:59 AM 18/02/2004 +0000, you wrote:
> >This may be a bug in IBO or FB 1.5 I am not sure,or it could be my
> >code,
>
> It's not so much your code (...hmmmm...) as your approach. From
what I can
> see, you are using a dataset component to perform a DML operation
(do-able
> but not recommended) AND you are kind of regurgitating it as some
kind of
> "second-level" dataset.
>