Allan McRae schrieb:
First off, I don't like modifying config files. But, given I did this update and still managed to screw my system up when testing it with a reboot...
So, the average advanced user won't even notice the problem, even you didn't (and you did get a .pacnew and a warning, didn't you?). It would take someone like me to notice it on his own - and let's admit it, there's not too many people like me.
Let's view this from another angle: It's not just the noobs and the unattentive users that will fall for this, it's also over half of the advanced and experienced users.
That said, we do modify configuration files all the time. We run grpck on a shadow update so users can still log in, some gtk update generate files in /etc so it still finds its plugins and more. We just don't do it ourselves, but hide behind some program provided to us and tell ourselves "It's okay, upstream wanted it this way". And guess what, nobody even notices.
So it is a question of which I hate more; post install messages or automatically fixing the file. A post install message means that I tell all complaining users that they should have read their pacman output. But hang on, they are already told that there is a .pacnew file for what is a very important config file.
Yes, and there has been a warning for inittab.pacnew several times in the past few months, always with some completely irrelevant added comment or added default lines. So, we give the user a pacnew with irrelevant things until he knows he can ignore it and THEN we break his system, how nice is that?
A short and simple message explaining what about this .pacnew is rather important might be in order. It could be as short as "Please read http://www.archlinux.org/news/1234 before you reboot." Or a bit more pragmatic, like "Your system is cannot reboot now. Please thank the Arch developers".
Attachment:
signature.asc
Description: OpenPGP digital signature
