forms a "backwards compatible" route to seeing cronjob output option parsing; tell when option set; writing stdout in as-it-happens or if-it-broke modes; tweak stalefile handling to assist following code;
I run it once a week, with a plain crontab line specifying two desktop machines to be seen fetching
when collecting %seen files =~ /^yyyymmdd-/, notice the longest (but don't do anything to either copy yet); when cleaning duplicate .tmp.file, unlink it if it is shorter (an old copy from fetch/expire race) but replace the done_file if it is longer (patch over our rename to .staletmp)
detect .tmp.DDD-TTT.RND files which are "old", but it's not obvious what to do with them (unless they have more info attached) so rename them for now; use @WARN rather than CORE::warn when things go wrong
If an entry gets too long, cut the middle part out. The beginning and end tend to be the interesting partsy anyway. In any case, you can click on the entry itself to read the full log.