Bug#473464: theoretical race that would result in missing menu items

2008-03-31 Thread Bill Allombert
On Sun, Mar 30, 2008 at 06:34:35PM -0400, Joey Hess wrote: Bill Allombert wrote: This is a known issue. A simpler way to trigger it is to run dpkg manually after apt. But since it does not result in two update-menus running at the same time, nor in a garbled menu, but only on a slighlty

Bug#473464: theoretical race that would result in missing menu items

2008-03-31 Thread Joey Hess
Bill Allombert wrote: Thanks you very much for the patch. Is there a timeline for dpkg triggers in unstable and testing ? The dpkg situation seems a bit, hum, confusing right now. I expect it will get in after it's had a while in experimental to work out any bugs. As I noted, the patch does

Bug#473464: theoretical race that would result in missing menu items

2008-03-30 Thread Joey Hess
Package: menu Version: 2.1.39 Severity: normal I think I've found a locking bug race in update-menus. This was found by inspection, I don't know how realistically it happens in the wild. Many apt-driven upgrades invoke dpkg multiple times. update-menus can be run inside each dpkg invocation. It

Bug#473464: theoretical race that would result in missing menu items

2008-03-30 Thread Bill Allombert
On Sun, Mar 30, 2008 at 03:56:18PM -0400, Joey Hess wrote: Package: menu Version: 2.1.39 Severity: normal I think I've found a locking bug race in update-menus. This was found by inspection, I don't know how realistically it happens in the wild. Many apt-driven upgrades invoke dpkg

Bug#473464: theoretical race that would result in missing menu items

2008-03-30 Thread Joey Hess
Bill Allombert wrote: This is a known issue. A simpler way to trigger it is to run dpkg manually after apt. But since it does not result in two update-menus running at the same time, nor in a garbled menu, but only on a slighlty outdated menu, it is sufficiently hard to trigger so that is not