Grouper Developers Forum

[Chris Hyzer <>]

1 in 9 quintillion?  So there *is* a chance?  I knew it.  :)

Also, it is sometimes useful to be able to run an update/insert query against 
the DB directly without having to worry about updating the hash col (or maybe 
people might not know about the hash col and they could make things 
inconsistent...)

Anyways, the other solutions are surely simpler, so does anyone need the 
constraint on all 3 cols instead of 2?
And (if there is a reason to have the constraint), does anyone need the 
default length of the col or be longer than 766?

Thanks
Chris

> -----Original Message-----
> From: David Langenberg 
> [mailto:]
> Sent: Thursday, March 06, 2008 2:04 PM
> To: Chris Hyzer
> Cc: Grouper Dev
> Subject: Re: [grouper-dev] changing lenght of attribute value column
>
>
> ----- "Chris Hyzer" 
> <>
>  wrote:
>
> > That is also an option, thanks.  I don’t see this issue in Jira, is
> > there an issue number?
> > If we really care that it is unique, this might not work since
> > multiple inputs can hash to the same value, so I would lean toward
> one
> > of the previous options... but if those aren’t ok, then this would be
> > better than now I guess
>
> Well, your constraint is on the triplet group_id, field_name,
> hash_value so collisions shouldn't happen anymore than they do now.  If
> you're referring to the MD5/SHA1 specific collision issues you're
> chances of a random collision (sha1) are about 1 in 9 quintillion.
> You've got a better chance of winning the lottery and being hit with
> lightning than getting a random collision.  Attacks are another thing,
> but honestly, if someone with malicious intent is in your RDBMS and
> able to exploit what should be a private, unexposed column by the API,
> via direct SQL you've got bigger problems.
>
> Dave
>
> --
> ***********************************************************
> David Langenberg
> Network Based Services
> The University of Chicago
> ***********************************************************