Re: When stability is pointless
Johannes Wiedersich wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi,
>
> Sam Kuper wrote:
>> 2008/11/5 Douglas A. Tutty <dtutty@vianet.ca>:
>>> Or, are you saying that you are trying to implement a psad recipe from
>>> the internet that doesn't apply to the version of psad supplied in
>>> Ubuntu?
>>
>> Essentially correct. But not just any old set of psad instructions:
>> the instructions provided on the psad website and in the developer's
>> book on Linux firewalls. In other words, pretty much the most
>> comprehensive set of instructions I could find.
>
> So what do you think that debian can do about it? For most packages I
> know, debian includes the correct version of the documentation. For git,
> as an example, the documentation at /usr/share/doc/ always corresponds
> to the version of git installed on the system and is upgraded along with
> git. (No need to search the web;-) ). Of course this is only possible,
> if the license of the documentation matches that of the software (ie. is
> free).
>
> If it is important for you for special cases, there is also
> backports.org from which you could install newer versions of certain
> software without compromising on stability for the rest of your system.
>
>>> For all Ubuntu is based on Debian, I don't think it follows debian
>>> policy. The policy manual says, basically and among other things, that
>>> installing a package should result in that package working
>>> out-of-the-box in some fashion only needing tweaking by the sysadmin.
>>
>> Define "working" (or "tweaking"). My experience with some packages in
>> Etch suggest that Debian sometimes has problems like this too.
>
> Just report a bug and the problem has a chance to get fixed for lenny.
>
>>> I've never used psad but I would be very surprised if the problem you
>>> experienced were to happen were you running Debian Stable.
>>
>> You may be right. Perhaps I should go back to Debian Stable. But one
>> of the reasons I switched to Ubuntu was to minimise the gap between a
>> package being deprecated by its developer and deprecated by its
>> maintainer, in an effort to avoid precisely the sort of problem I
>> outlined in my post.
>
> I think this shows the point where you misunderstand how debian works:
>
> There are three levels any package can reach:
>
> - unstable/sid: frequently updated from upstream, latest software
>
> - testing: software has been tested some time and should contain less
> changes and bugs than unstable
>
> - stable: software has been extensively tested to work. It is rather
> unfrequently updated (about 1.5 years between releases) and hence you
> get a 'stable' system to work with.
>
> Pick whichever suits you. You obviously can't have stable software and
> frequent updates at the same time...
>
> It is also impossible to predict the right level of stability for
> everyone and for every package...
>
> ... or to predict which version of a package will be featured in a book
> or in some other documentation...
>
> ;-(
>
>>> Since Ubuntu is based not on Debian Stable but on (I think) Unstable, I
>>> don't know how one can consider any Ubuntu release to be stable.
>>
>> Ubuntu has LTS (Long-Term Support) releases, which roughly translate to
>> Stable.
>
> Yes, but IIRC it is still based on debian sid. Ie. it never transitioned
> debians unstable - testing - stable queue. IIRC it just means that the
> developers made a commitment to extend security support. (I hope someone
> will correct me, if I'm wrong)
>
Sorry to mix up in your discussion, but I hope sharing my experience will
benefit you.
I've been using debian since potato and must say that it has a genius
versioning system. Actually there are 4 levels for packages, you've
missed "experimental".
So I do use the stable version on my servers and the testing on my notebook
or desktop. It works just great. I should admit that k/ubuntu has much
better configurable or configured desktop settings, but being derived from
unstable it has still problems, though the programmers/developers there
swear that their software (k/ubuntu) is fine.
I've tested kubuntu for over a month now and there are such small problems
that finally broth me back to lenny for desktop. I think the ubuntu problem
is really deriving it from unstable, but I hope they know what they are
doing. From my point of view this is simply not relyable for production use
I mean for getting a job done. I spent about 2 days to make it work for me
and still there are some things that do not work like they should. So I
don't see any difference between lenny (debian testing) and k/ubuntu.
>From my point of view debian testing has the advantage, that at some point
in time it becomes stable which means that you can use this for other few
years and don't have to change anything.
If you miss a package, like I do with few packages, I download the code,
compile, install, test, debug until it's working. then I make a deb
package, install and forget about having troubles with it. It's just great.
regards
Reply to: