2009/2/8 Cristian Rodriguez
Rob OpenSuSE escribió:
2009/2/7 Cristian Rodriguez
:
The user base, don't think "zypper dup" is a supported upgrade method from 11.0 to 11.1, nor and there is a fate entry on supporting a "Debian like dist-upgradee".
It can be supported, but when using a limited set of known repositories, IMNSHO.
Sounds like a very good start.
The fate entry on the subject.
FATE cannot contain everything that has to be done or cleaned up before this to work, you can write a book of the size of "Don Quixote" on the subject, for the user is just zypper dup, internally for developers is a great deal of work, you have to deal with among other things
- API issues - ABI issues ( tricky as hell, hint hint) - Correct a huge set of packaging problems - track backward incompatible configuration changes, without destroying user's configuration - Debug and fix all the bugs introduced by this changes (and goto 1)
So far, no one has done it right...;-)
The Debian solution, which is generally praised, doesn't attempt to solve all of these things. They seem to have managed to make it work "well enough" rather than "bullet proof", and not guaranteed dist-upgrade (full-update) whilst running GUI for example. The "Debug & fix" point, is why it needs clear publicity and testing effort. People can plan test areas, copy their configurations across and try updating, if they know it is supported. If API & ABI issues are a problem, then statically linked tools that update files on disk from one consistent state to another, would seem to be a good solution. Disk formats could change to (like when RAID & LVM formats changed), but some documented non-live procedure, to upgrade deprecated features would likely solve it for most ppl.
After 2 intervening Ubuntu & Fedora releases 11.1 will look stale.
why it will look stale ? we are already making a new product , it takes time and considerable effort !
Simply because Fedora & Ubuntu will be re-freshed more often. For example 11.1 has KDE 4.1 + backports, which is not what real end users want, they actually want KDE 4.2 or 3.5.10 now, and 11.1 is still going to install a reviled KDE4 software version (I know that's unfair but it is the general perception). So the point of the thread was to discuss alternatives. It has been discussed (somewhat) before around Christmas time. Those making suggestions are looking for solutions : - perceived shortage of alpha & beta testers - perceived reliability & installation success - perceived drawback of tying openSUSE release to other projects - difficulty verifying bug fixes (they are 'done' but checking that, means upgrading to a release that may not be available till months later) Really the roughly 3 releases in 2 years, talked about is doing same thing as the other distro's just slower. As I have good memory of pain of installing 10.3, and time it took to stabilise on my machines (yes I was unlucky), I do not have faith that simply extending the time till next release, will really change anything. The move to make Factory updates more efficient, does seem like a step to improve the situation. The easier it is to get close to development, the greater number (in time) of useful tester and developer contributions.
so what ? you want a perfect OS ?.. Im sorry to bring you down to earth, there is no such thing and will most likely never be, people will always whine and complain anyway ..it is very easy to do so when you are not one doing the work...
Actually I agree, that does mean that it's wrong to look to improve the situation. As it stands, being an active openSUSE community member, testing the alpha's & beta's, is a fair amount of work, reported bugs don't necessarily get fixed, nor were work rounds published in release notes. When bugs are reported, sometimes it appears that the assigned engineers are under pressure to get them CLOSED or RESOLVED in some quick way, rather than really solved.
and also planning to re-issue ISO's would hopefully help those who found 11.1 wouldn't boot on their machines.
So you want people to waste the already limited resources available to respin a release just due to "being unable to boot on certain specific machines" ?? Im sorry but that does not qualify as a good reason for me.
11.1 failing to install on an 11.0 machine looks like a significant regression to me, and not a "waste" to resolve. If sorting that problem involves "waste of limited resources" then there's something very wrong in the release process. As it stands, it appears that the release mentality required for retail sales of box sets is entrenched. openSUSE users are not the types needing certification and QA approval processes. Those with good net connections, really want the best possible OS, more of the time, with different compromises on new feature / stability trade off. Those without decent net connections, need solid releases, not stuff that need lots of updates, or ones that don't boot. All the big community releases, not tied by policies of a commercial distributer, have gone for continuous update models, with periodic release of installation media. There's a reason for that, and is not because they didn't listen to end-user requirements, or were not dog fooding. -- To unsubscribe, e-mail: opensuse-project+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-project+help@opensuse.org