> 
> [...]
> > Well, both, though you are right, my first thoughts were for the use of
suExec for the 
> > WebGUI.
> >  
> The WebGUI is actually doing two things:
> 1) reading data from DSPAM_HOME
> 2) executing the dspam binary
> 
> To avoid suExec for (1) you need to have ALL data in the database. Not
only quarantine.
> To avoid suExec for (2) you could add the web server user to be a trusted
user.
> 
> The WebGUI however would need to be much more complex if everything is in
a database. The UI would need to implement direct access logic to the
database. Right now the WebGUI is storage backend unaware.
> 

Both solutions doest feel right :-)
Why not everything in the database? 
Ok maybe not possible to run every currently support DB Backend. But again ,
DSPAM is Design as an Enterprisesystem not as an Clientsystem where i have
to drive lowcost

BTW  Big Quarantines would be processed much faster if everything is in the
DB. YOu do not have to parse a hole 200MB file

And i CANNOT agree thats much more complex fort he webui - i tell you its
defenetly easier since you do not have to parse the files and bother with
all those problems comes with it.
And more... you can split the mails on different tables
(attachment/subject/content) this should be done by dspam
TO Read you do not have to parse and filter (where always can go something
wrong o rat least tage ages) 
You simply request the needed information

For example: to show the whole list of messages stored in a mysql DB its one
sql is  5 rows of code (in the very simply variant :-)
Where the problem with direct DB access? Perl/php/java have it all built in 

And you do not need anything. You can run in jail or whatever you want
(another thing i dont like apache being not jailed) no suexec or another
thing with users and rights


Simply DSPAM gets a message / decide to quarantine it /  split and write it
to the DB 


>From there simply fetch it and show it.


You do not need additional utilitys to delete very old quarantines - one
statement is good for it


The Filesystem version isnt simplier - its much more complicated - we nee
down atmic here (locks and co) 
We need to parse and bother with charset
We need to bother with user UIDs, setuids, and so on
We need third party tools to clean
We do not have a real overview how those files grow

User cannot set their own groups
We cannot autoassign by vertain rules since that would be more complicated

And so on .....

There are so many things possible to do much easier / or even possible to do
by storing everything in the DB backend 




PS do you think google saves their images on a filesystem or in a database
... hmm guess :-)


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

_______________________________________________
Dspam-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dspam-user

Reply via email to