[Grub-dev] managing grub downloads the wiki way

[jer]

I've thought about this quite a bit and I think there's a much  
cleaner way to manage access to sensitive resources (like bulk  
compressed up-to-date crawl snapshot, and in the future, indexes).

Instead of reporting transfers and basing access on wiki accounts,  
instead there is simply one wiki page that has an IP access list in  
it where each list item is an IP address, full or partial.  There  
would be a script that is triggered on edits to pull the IPs out and  
put them in a .htaccess file, allowing only those IPs to connect to  
access the resources.  It's very clean, very logical, and easy to  
watch/manage.

I'll try to set up an experiment with this method soon and the  
current grub data so we can see how it feels in practice...

Jer

On Oct 9, 2007, at 11:19 AM, jer wrote:

> If this has been tried or is in use elsewhere I'd love to learn  
> more, but we have a need to let the community openly manage access  
> to the resources we're jointly building, namely the source crawl  
> data and resulting index.
>
> I'm approaching this very simply, that there would be a site like  
> cp.grub.org that would only host the bulk data sources and you'd  
> need to authenticate using your wiki username.  The web server  
> would simply log how much you transferred, and these logs would be  
> aggregated into a few special reports that anyone can keep an eye  
> on.  Any user could be blocked from just the cp.grub.org and would  
> have to make a case (on their talk page?) if they feel they  
> shouldn't be.
>
> We'd need to develop a good social policy about handling these  
> functions, at what point is someone downloading too much, what kind  
> of reasons are there for needing access, etc.  In the beginning the  
> data will be rather small and unimportant, but as it grows so will  
> the need to be more and more careful, so I want to make sure  
> everyone starts thinking about this process ahead of time.
>
> Jer
>
>