I had taken one shortcut in relying on the assumption that I would get unique
package names from
available.packages(), and that seemingly was
not the case. I added another lookup in the sql backend to the script: for
every possible new package, we explicitly test whether the package and
version combination exists or not. This will prevent the 'cairoDevice package
loop' from showing a few more iterations. Sorry for the useless entries, but
we should be back in good service now.
However, yesterday's initial 'flood' a new posts appears to have triggered a 'flood prevention' mechanism in the Planet agggregators and/or Google Reader. Looking at my input directory, I noticed over sixty new entries between 'new' and 'updated' entries :
edd@ron:~> find cranberries/bloginput/cran/ -type f -name \*txt -ctime -3 -cmin +1200 | wc -l 63 edd@ron:~>
However, Google Reader only showed me 40 posts. So if you're a keen consumer
of CRANberries, and intent of catching all news (yes, I'm talking to both of
Oh, and if anyone knows how to fix lvm partition info, I'd greatly appreciate an email at the usual address.