On Tue, Feb 12, 2013 at 09:55:12PM +0000, Reuben Thomas wrote: > I was interested in the definition of the C main function, so I > hopefully typed "man main". Well, it turns out this function doesn't > have a man page, but I was surprised to see that I got a page for > mpd-dynamic(1PERL), a page which doesn't even contain the characters > "main". > > Here's the output of "man --debug main"; I confess I'm no wiser after > reading it, though I can see where it somehow makes the connection > between the word "main" and the above-mentioned man page. > > I'm not at all sure there is a bug here, but it'd be nice to > understand what is going on, at least!
That's thoroughly weird, and having looked at the page I'm not immediately any the wiser either. I wonder if perhaps this is some revenant of a previous version which had a whatis reference for "main" in that page. So let's dig: > name: main > sec. ext: 1p > section: 1 > comp. ext: gz > id: C > st_mtime 1318989731 > pointer: mpd-dynamic > filter: - > whatis: Your reportbug metadata suggests you're using Ubuntu 12.10, which had libaudio-mpd-perl 1.120610 (the current version). The timestamp on /usr/share/man/man1/mpd-dynamic.1p.gz there is 1335936468, a.k.a. 2012-05-02 05:27:48 UTC; that reference is dated 2011-10-19 02:02:11 UTC. That disparity is suspicious, because st_mtime should always have been an mtime picked up from statting a page on disk. Looking through https://launchpad.net/ubuntu/+source/libaudio-mpd-perl/+publishinghistory I notice that it's a day and a half after version 1.112670-1 entered Ubuntu 12.04, and since that was a new series it would have taken a while for the sudden flood of updates to finish building. Following links to the build log at https://launchpadlibrarian.net/83150624/buildlog_ubuntu-precise-i386.libaudio-mpd-perl_1.112670-1_BUILDING.txt.gz, indeed your timestamp matches when that version was built to within a minute. And if I look at bin/mpd-dynamic in that version of the source package I see: =head1 NAME main - a dynamic playlist for mpd So there's our smoking gun. The question is then why mandb didn't notice that that was a stale reference and remove it after upgrading to a version with a more sensible NAME section, and at this point I've not had time to refresh my memory well enough to determine whether this is a missing feature in mandb or a bug; I will return to this bug as soon as I can manage it and address that question. At least now I should be able to distil a test case from this. In the meantime, it would possibly save me time going down a blind alley if you could confirm whether /etc/cron.weekly/man-db is active on your system, and that it does not use the -p or --no-purge options when invoking mandb. Thanks, -- Colin Watson [[email protected]] -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
