Re: [Rule-list] test reports and benchmarks

[Martin Stricker]

"Rodolfo J. Paiz" wrote:

> One: sadly, there is always out there some bastard who will try to
> screw with your results just to see if it can be done, or to muck up
> your database.

Sad but true.

> I suggest we allow only people who have registered Test PC's to submit
> such input, and check that the report submitted matches information
> they have provided previously such as CPU type, RAM installed, or
> other. I also suggest that the program somehow encrypt this file so it
> cannot be altered after generation... just so someone doesn't someday
> try to make his/her box look "faster".

Even better, include an additional field into the database: Status,
having values of "submitted" (s) and "visible" (v). The first is the
default assigned to every new test record. The second will be assigned
by the RULE test team if the record is considered useful (see next
paragraph). In any reports or views only records with status='v' will be
seen. Thus the bad records remain in the database, making it easier to
recognize people who try to cheat again.

The database should have set the privileges accordingly: All people with
a registered test PC are GRANTed insert privileges, but only the RULE
test team (Benoit and maybe others) have update privileges. Nothing
should be deleted, you never know if you can use the data later on.

Best regards,
Martin Stricker
-- 
Homepage: http://www.martin-stricker.de/
Webmaster-Forum: http://www.masterportal24.com/cgi-bin/YaBB.cgi
Red Hat Linux 7.2 for low memory: http://www.rule-project.org/rule/
Registered Linux user #210635: http://counter.li.org/