[opencms-dev] Pluggable User Modules
Michael Goodwin
mgoodwin at amadeus.net
Wed May 26 17:37:01 CEST 2004
Thanks for the quick response,
>The name of the driver packages is not hardcoded. I know since we have
>already developed drivers that use a com.alkacon.* package structure.
Sorry, yes I realise now. The class comes later in opencms.properties. The
driver.user specifies a category. From looking at the code it appears that
multiple drivers are usable as a failover list , (but the failover is not
implemented so far anyway in the generic sql drivers).
>Ok I just removed that "final" from the methods in question (actually
there
>where only 5 of them). AFAIK the assumtion that "final" allows special
>optimization is not true anymore for modern hotspot JVMs anyway.
Cool
>I don't fully understand what you want here. Can you post me some code
>sample (better patch) with the required changes?
No problem, be tomorrow tho.
Thanks,
Mike.
From: "Alexander Kandzior" <alex at opencms.org>@opencms.org on 26/05/2004
17:05 ZE2
Please respond to opencms-dev at opencms.org
Sent by: opencms-dev-admin at opencms.org
To: <opencms-dev at opencms.org>
cc:
Subjec RE: [opencms-dev] Pluggable User
t: Modules
Michael,
regarding your comments to CVS HEAD:
> *The 'configurability' of the
> plugin appears to be limited to the namespace
> org.opencms.[choose a name] in opencms.properties. looking at
> the line (from my memory something like):- driver.user=db
>
> The driver I'm going to create is specific to our needs and
> does not belong in this package namespace (and I assume this
> will be true for most UserDrivers that people will write).
The name of the driver packages is not hardcoded. I know since we have
already developed drivers that use a com.alkacon.* package structure.
>
> *Some of the methods in the generic sql UserDriver are final
> (with the reason cited being that it allows the compiler to
> inline). This may be a case of premature optimisation. The
> method is long, and it makes subclassing much less easy/useful.
Ok I just removed that "final" from the methods in question (actually there
where only 5 of them). AFAIK the assumtion that "final" allows special
optimization is not true anymore for modern hotspot JVMs anyway.
> *Additional user info is stored as a serialized blob in the
> user table. Null values cannot be read in from the database.
> Due to trying to aggregate all users together into one
> database, a number of fields will not apply to most users and
> I think it would be good to be able to leave them null
> (especially for larger fields).
I don't fully understand what you want here. Can you post me some code
sample (better patch) with the required changes?
Best Regards,
Alex.
Alexander Kandzior
Alkacon Software - The OpenCms Experts
http://www.alkacon.com
_______________________________________________
This mail is send to you from the opencms-dev mailing list
To change your list options, or to unsubscribe from the list, please visit
http://mail.opencms.org/mailman/listinfo/opencms-dev
More information about the opencms-dev
mailing list