doktor5000 wrote:On the other hand, if you want to skip so much packages, that's not a problem of urpmi or skip.list, but it's rather your custom-made problem.
We'll have to agree to disagree on that. I had no part in creating the imagined dependencies for the two packages which cause the additional 145 files to flagged as required. I say the dependencies are
imagined because in English a dependency, something on which another depends, implies some form of functional requirement without which the thing which does the depending cannot exist/function/work as intended. In the context of software packaging one may easily imagine that a packaged application may depend on a library which it uses. One would not easily understand why it might
depend on a library which it does not use, and still less easily might one understand why an application could possibly depend on some other application, the function, purpose and even existence of which is completely unknown to the user.
In the given case I am using LXDE with a small number of KDE applications which are too useful/familiar to leave out. I expect that when I installed those KDE programs they very reasonably caused the
Default-kde4-config and
mageia-kde4-config-common packages to be installed as dependencies. After all, some programs need some config. But now the cart has been put before the horse. Now the updated versions of these config packages
depend on a small universe of applications and their libraries!
Why do I care? I can assure you it isn't just pure bloody-mindedness. I may be alone in this, but I care about security, economy, efficiency and fitness for purpose. This example of the growing tendency to subvert the main purpose of the rpm dependency impacts on each of these goals.
Security
In another context it would be regarded as a security weakness to tolerate an attack vector which allows programs not chosen by the user to be installed and run. This is what some Mandriva (and now, apparently, Mageia) packagers have done when they have recklessly expanded the definition of what is required to include whatever is convenient for them.
Economy, Efficiency & fitness for purpose
I have elected to use a small, easily transportable external disc drive for Mageia 2. It is exactly big enough for my needs. It can travel with me and provide just what I need wherever I may be. As I cannot always expect to have a high speed internet connection available, or a high speed computer to boot it on, or even a mains electricity supply to power it, it is important to me that I engineer its facilities as carefully as possible to maximise its efficiency. I spend time selecting the tools which I will install to ensure that I have a system which is pleasant to use with the smallest possible footprint on the limited available disc space. This produces further advantages in reducing the bandwidth and time needed for security and other software updates to the minimum.
Causing config files to
depend on their applications is just wrong. The real dependency is the other way round. Coincidentally, the real dependency is also more efficient, and more secure (by my definitions). I will probably never convince anyone of this so I must find the tools to allow me to accept this less-than-perfect situation and work around it. It is only my "custom made problem" because I prefer economical, efficient and secure to extravagant, expedient and sloppy.
On a lighter note, I disagree with your implied assertion that urpmi and the skip.list file are not the proper tools for this package management problem of mine. Just because it is awkward and time consuming to set up it doesn't mean that urpmi isn't a great package management tool. I agree that it would be almost perfect if the expertise to create regular expressions for the skip list were included in rpmdrake so that a simple tick in the box would build your list and grey-out the entry for the package in the updates list. It is much easier to tick 147 boxes than write them out by hand.
Richard