On Thu, Dec 12, 2002 at 03:26:30AM +0000, Oliver Elphick scribbled: [snip] > > > It would take days to test all possible paths properly. I don't have > > > time, and I don't have any volunteers lining up to help either. I check > > > that it installs for me. As for other possibilities - that's what > > > unstable is for. Sorry. > > No, ustable is not for breaking people's work machines, sorry. If you aren't > > sure of your package, upload it to experimental, please. If you want to > > experimental just doesn't get enough exposure. I'm aware of only one > person who downloaded the initial packages that I put there. Not enough publicity? > > upload to unstable, you are taking responsibility for the systems of > > hundreds of developers out there and if you fail to test trivial cases like > > the one above, you are abusing their confidence. This is not a good thing > > If you are running your work machine on unstable, you are taking a lot > of risks; too many by the sound of it. I won't use sarge for my work, it's too old for me. But I do not keep any important databases on my machine at home. And, as I said, loss of data wasn't my concern. The attitude was (still is). > http://www.debian.org/distrib/packages says: > Packages in unstable are the least tested and may contain > problems severe enough to affect the stability of your system. > Only experienced users should consider using this distribution. > > To put it another way, if it breaks, you get to keep all the pieces. Sure, but there are different kinds of breakage. Those resulting from lack of testing (basic testing) aren't excusable. Never. > You can confidently expect to have serious problems once or twice a year > if you track unstable. That's why we have unstable: so that less When potato was unstable I remember only one serious problem - bash. Since the release of Woody, there have been many problems with GNOME, quite a few with your postgres (the recent 7.3 releases are particularily diastrous - there wasn't a single update including the current one - 7.3-6 - which went smoothly.) > competent users aren't exposed to those problems. You seem to want the > quality of stable out of unstable, which is not reasonable. Debian's > collaborative effort includes package testing; that is what unstable and > testing are for. Thank you for participating, however unwillingly. You completely misunderstood my mail. The point is not that I'm complaining about bugs. I am competent enough to cope with them. I'm competent enough to cope with the kind of screwups in bonobo, mozilla or postgres. Don't worry about it. And I do not want rock solid unstable, I just want people to test what they do. You added the postgresql.env code in March and you have never tested it - that's unresponsible and reckless. Mozilla doesn't install because of a incorrect command syntax. console-tools screw up people's terminals under X, because the package wasn't tested for that. bonobo-activation prevents people from logging into GNOME, because it was never tested (installed in this case). That's what I'm worrying about. We can have two dozens of bugs a day in unstable (I hope not, though :)) as long as they are "legitimate" bugs and not silly errors resulting from laziness, lack of time, lack of will or anything else equally unacceptable for a professional developer. > > and 'unstable' doesn't mean 'broken'. Last version of postgresql-dev had a > > broken .postrm script - if you remove a file which you aren't sure whether > > it exists or not, it is nice to do > > Packaging changes have consequences, some unforeseen. That was a Come on! That error would have came up if you had tested it. > hangover from an old version of the package, which became invalid > following upstream changes and extensive repackaging. If I were And the hangover could have been cured in the next version. It wasn't. > perfect, I would have noticed before other people told me; but on my > system it didn't happen, so I didn't notice. See? That's what I mean - you didn't test it enough. You are maintaining a crucial package, you are expected to take more steps in order to assert the package quality than most other developers. Cope with it or ask for help - or quit. > [Copying this to debian-devel, rather than the bugs database; since > others may wish to comment.] Yep, sure. thanks, marek
Attachment:
pgpR6e9W5uz5e.pgp
Description: PGP signature