diff options
Diffstat (limited to 'zarb-ml/mageia-discuss/20100929/001513.html')
| -rw-r--r-- | zarb-ml/mageia-discuss/20100929/001513.html | 158 |
1 files changed, 158 insertions, 0 deletions
diff --git a/zarb-ml/mageia-discuss/20100929/001513.html b/zarb-ml/mageia-discuss/20100929/001513.html new file mode 100644 index 000000000..abef9f812 --- /dev/null +++ b/zarb-ml/mageia-discuss/20100929/001513.html @@ -0,0 +1,158 @@ +<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN"> +<HTML> + <HEAD> + <TITLE> [Mageia-discuss] Package management system + </TITLE> + <LINK REL="Index" HREF="index.html" > + <LINK REL="made" HREF="mailto:mageia-discuss%40mageia.org?Subject=Re%3A%20%5BMageia-discuss%5D%20Package%20management%20system&In-Reply-To=%3C1285713656.3727.502.camel%40akroma.ephaone.org%3E"> + <META NAME="robots" CONTENT="index,nofollow"> + <META http-equiv="Content-Type" content="text/html; charset=us-ascii"> + <LINK REL="Previous" HREF="001512.html"> + <LINK REL="Next" HREF="001518.html"> + </HEAD> + <BODY BGCOLOR="#ffffff"> + <H1>[Mageia-discuss] Package management system</H1> + <B>Michael Scherer</B> + <A HREF="mailto:mageia-discuss%40mageia.org?Subject=Re%3A%20%5BMageia-discuss%5D%20Package%20management%20system&In-Reply-To=%3C1285713656.3727.502.camel%40akroma.ephaone.org%3E" + TITLE="[Mageia-discuss] Package management system">misc at zarb.org + </A><BR> + <I>Wed Sep 29 00:40:56 CEST 2010</I> + <P><UL> + <LI>Previous message: <A HREF="001512.html">[Mageia-discuss] Package management system +</A></li> + <LI>Next message: <A HREF="001518.html">[Mageia-discuss] Package management system +</A></li> + <LI> <B>Messages sorted by:</B> + <a href="date.html#1513">[ date ]</a> + <a href="thread.html#1513">[ thread ]</a> + <a href="subject.html#1513">[ subject ]</a> + <a href="author.html#1513">[ author ]</a> + </LI> + </UL> + <HR> +<!--beginarticle--> +<PRE>Le mardi 28 septembre 2010 à 23:16 +0100, Richard a écrit : +><i> On Tuesday 28 September 2010 22:21:08 Michael Scherer wrote: +</I>><i> > Le mardi 28 septembre 2010 à 20:20 +0100, Richard a écrit : +</I>><i> > > How much better could it possibly do this? What am I missing? You have +</I>><i> > > both mentioned alternatives, some of which I know by name, but in what +</I>><i> > > way do apt, yum or smart do this job any better? +</I>><i> > +</I>><i> > Well, apt is likely to be faster, c++ may be the cause. +</I>><i> > +</I>><i> +</I>><i> That makes sudden sense; I had forgotten how slow rpmdrake has become while it +</I>><i> builds and rebuilds its view of the installed packages inventory. I am +</I>><i> guessing this is a "compiled-v-interpreted" thing. No matter how much faster +</I>><i> is the PC I use, rpmdrake always seems to be slower than I would like/expect. +</I>><i> That, of course, is where urpmi is most certainly your friend being many many +</I>><i> times faster to get going - provided I already know the name of what I want +</I>><i> to install. +</I> +Well, I think the rise of number packages, and the time needed to create +graphical objects are the reason of the slowdown. Maybe there is also a +regression somewhere. But you cannot really compare apt and rpmdrake, +apt being command line only, there is no graphical interface creation +overhead. + +><i> > Smart is portable across type of repository. It also use a cleaner +</I>><i> > design or algorithm, according to his developer. Among nice features, it +</I>><i> > can draw graphs of the dependency, feature a command line shell or +</I>><i> > parallel downloads ( <A HREF="http://labix.org/smart/features">http://labix.org/smart/features</A> ) +</I>><i> > +</I>><i> A dependency graph would be pretty to look at, but it would be even better if +</I>><i> the dependencies were to be colour coded to indicate which are true +</I>><i> dependencies and which are just the packager's whim, and then give you the +</I>><i> opportunity to ignore those which are neither wanted nor needed. +</I> +In rpm, all deps are usually required. You can have Suggests, but +Suggests can be removed without troubles. + + +><i> > > I realise that package managers are needed because humans have to add +</I>><i> > > some intelligence in the form of what libraries are needed to get a +</I>><i> > > program to run. I also know that sometimes humans get this badly wrong +</I>><i> > > (try removing a library that you know will never be used and ask yourself +</I>><i> > > why rpmdrake wants to remove over 200 packages with it!). Do other +</I>><i> > > package managers manage to avoid this embarassing and frustrating +</I>><i> > > behaviour? or is it that it is just easier to get it right with package +</I>><i> > > types other than rpm? +</I>><i> > +</I>><i> > Nope, the problem is not linked to rpm or deb. If a library is needed, +</I>><i> > it is needed, that's simple. +</I>><i> > +</I>><i> > A system like emerge or macports ( macports is also in contribs, afaik ) +</I>><i> > may however reduce the required dependency, depending on the software. +</I>><i> > +</I>><i> I sense a tendency in your response to the same confusion I am suffering from. +</I>><i> On the one hand I find it easy to imagine that the packager is in total +</I>><i> control of what a package regards as a "dependency" but on the other hand I +</I>><i> must accept that your comment on macports may also be true. I do not know yet +</I>><i> how to reconcile these apparent contradictions and I am really hoping I don't +</I>><i> have to become the master of package management on all conceivable platforms +</I>><i> before I can understand enough of this to make sense of it. +</I>><i> +</I>><i> I can fully grasp the concept that the structure of a package (rpm or deb) is +</I>><i> not relevant to the program's requirement for one or more libraries, but I do +</I>><i> not understand by what mechanism "nice to haves" are included as "must +</I>><i> haves". Is it likely to be simply a packager's oversight or perhaps do some +</I>><i> package types not enable the distinction to be made? If the latter, and rpms +</I>><i> are of this type, then is ther a package type which does support the +</I>><i> distinction? +</I> +Macports and emerge are more flexible because they compile packages. + +Ie, some packages are inflexible. They requires strict dependency. +For exemple, mplayer requires glibc, and this cannot be changed. + +Some have compile time option ( --with, --without at ./configure ). + +Mplayer can be compiled without libvorbis support, but you need to +recompile it to enable it if needed later. This is quite annoying for +most users, so usually, the packager have to choose, and in mandriva, we +usually enable this, as most people will prefer features, and those that +have specific requirement are usually able to fix the issue themselves +( or would use another distro like gentoo, we cannot target every +possible use case ) + +And some have runtime options ( ie a plugins system ). Mplayer is not +like that, but totem ( based on gstreamer ) is. Ie, you can install +totem, and it will not automatically pull every gstreamer plugin to read +every file. + +Packager usually try to split plugin in separate rpms, but then you have +to make a choice, ie do we want the plugin to be installed by default or +not ? It is a packager choice usually. + +If you fill there is excessive requirement on a package, feel free to +ask the packager his opinion, or open a bug. But I would recommend that +you first take some time to understand how it work before opening lots +of bugs regarding this . + +-- +Michael Scherer + +</PRE> + + + +<!--endarticle--> + <HR> + <P><UL> + <!--threads--> + <LI>Previous message: <A HREF="001512.html">[Mageia-discuss] Package management system +</A></li> + <LI>Next message: <A HREF="001518.html">[Mageia-discuss] Package management system +</A></li> + <LI> <B>Messages sorted by:</B> + <a href="date.html#1513">[ date ]</a> + <a href="thread.html#1513">[ thread ]</a> + <a href="subject.html#1513">[ subject ]</a> + <a href="author.html#1513">[ author ]</a> + </LI> + </UL> + +<hr> +<a href="https://www.mageia.org/mailman/listinfo/mageia-discuss">More information about the Mageia-discuss +mailing list</a><br> +</body></html> |