Hi all,

First of all, I love Arch so much, and almost every person to whom I
presented it switched. However, like other people here, I've been
annoyed by problems recently.

Arch used to be VERY stable AND bleeding edge. I don't want to put the
fault on anybody, and I'm not following the lists or forum anyway to
know who's fault is that.

Anyway, I think Arch has the potential to be both a bleeding edge
distro, and a production compatible distrib.

About updates, I generally assumes that things will work fine, and not
the contrary. Thus I don't go check every single page of the internet
to find out if my daily update will break my laptop. So, I'd rather
have a ML announce.

Now, the world is not perfect, and I don't expect Arch to be rock
stable, that's bleeding edge. However, I think that there are two
simple way to improve testing and avoid breaking:

1. Put more notices when installing packages. If there is an issue
when upgrading foo.x to foo.y, pacman should tell me when I try to
upgrade. This is the ONLY place where I'm sure I will get the notice
(maybe I don't read email, nor RSS reader, nor whatever else)

2. Use unionfs (provided in kernel26archck) when upgrading. I never
used it myself, so I'll give an example from what I've understood of
if. You want to upgrade your box:
- you run a 'pre-script' that uses unionfs to protect all important
directories (bin, sbin, usr, opt, etc...) by mounting them read-only
with changes put in /changes
- you upgrade

- Either everything went fine and I guess you can umount and put back
the changes from /changes to the real root
- Either something broke, you umount and your system is clean.

That would be a nice way to work, avoid the mess of finding packages
to downgrade, and would even allow more people to use testing, if they
know they're not taking too much risks.

What do you think ?

_______________________________________________
arch mailing list
[email protected]
http://www.archlinux.org/mailman/listinfo/arch

Reply via email to