On Wednesday 31 July 2013 04:42:29 Marguerite Su wrote:
On Wed, Jul 31, 2013 at 4:37 AM, Stephan Kulow
wrote: d) Too few people actually care and fix problems of importance
Actually nowadays, after submitting a package to our Factory, devel repository maintainers and package maintainers don't even know there're problems at all, until we want to update the package next time.
Because when you submit a package to Factory, it of course should be built against current Factory. So we thought it builds.
Then future change makes it broken. But how do we know? eg: I declare, well, no one knows what packages are broken with the lastest gcc 4.8, even after coolo actually pushed 4.8 into Factory.
Actually few repo maintainers really read the "50 failures" status of repository every day or week. few package maintainers really read the Factory status of her package every day or week.
So actually she is in a black box until next time she opens the package view page.
Coolo has made some improvements in the past: he emails us/our mailing lists.
But a lot of packagers are just hit-and-run people. They don't even join the ML. They don't even read coolo's email. They just think: anyway I submit the package, it should be your responsibility to maintain it.
That's bad. We're the distribution which has the most packages, while we're also the distribution which has the most unmaintained packages.
Perhaps, at least, we could improve the integrated automated tests to make sure that pushing things that break other things will simply be impossible...
Greetings
Marguerite