Re: [Savannah-hackers-public] Savannah recovery TODO

[Sylvain Beucler]

Hi Karl,

The plan is to re-checkout everything from May 27th (which Ward is
currently doing) and then provide a diff for projects to recommit.

Sounds ok?

-- 
Sylvain

On Wed, Jun 03, 2009 at 04:18:01PM -0500, Karl Berry wrote:
>     - Decide if restore webcvs from May 27th or attempt to recover the
>       last 2 days commits from www.gnu.org and www.nongnu.org; 
> 
> Would it be possible to detect which webcvs's were modified after the
> May 27 backup?  I'm thinking of something like a find-by-time(*) on the
> live repos and just look for files which were modified.
> 
> I expect there are very few projects which had web changes in that short
> span.  If we could explicitly tell the maintainers of them to deal with
> their webcvs, that might end up taking considerably less time than a
> heroic effort to detect and recover every project.
> 
> Just an idea.
> 
> Exception: speaking with my gnu webmaster hat on, it would be highly
> desirable to bring the www CVS itself up to date.
> 
> Thanks,
> Karl
> 
> (*) I'm sure most people here have their own methods for this, but just
> in case it helps, here is my little script to output all files sorted by
> mtime (also shows file size).
> 
> find "$@" -xdev -path /proc -prune -o -type f -printf "%T@ %k %p\n" \
> | sort -nr \
> | gawk '{ T=$1; k=$2; $1=""; $2=""; sub (/^ */, "", $0);
>           printf "%s %5d %s\n", strftime ("%Y%m%d.%H%M", T), k, $0}'
>