1 |
You're absoutely right. The LSB guys aren't going to provide us with |
2 |
'software map' files or RPMs themselves. The only people likely to do that |
3 |
are the maintainers of a distro or, the developers of the program who would |
4 |
then lean (naturally) to their own preferred distro. |
5 |
|
6 |
But why is it that whenever I want to download some package from its |
7 |
home site I see only one source tarball but more RPMs than I can count? Is |
8 |
this the 'standard' they propose? Imagine a caricature: the LSB stands |
9 |
between Redhat, SuSE and Mandrake, planting its flag with the motto: the |
10 |
center of the earth is Here. Gentoo and some other stragglers are seen on the |
11 |
horizon. |
12 |
|
13 |
Instead of _making_ a standard, they _selected_ one. Instead of reconciling |
14 |
the differences between the distributions, they've selected a feature which |
15 |
several have and the others deprecate and said: this is Right. It is a Good |
16 |
Thing. I call this monopoly practice, and discouraging competion. I could |
17 |
almost believe Redhat bribed the LSB. Ugh. |
18 |
|
19 |
If pursued, this policy (not only with regard to RPMs, but other similar |
20 |
propositions as well) won't 'standardize' and 'unite' anyone - it can only |
21 |
create a rift between the RPM-based distros and those that aren't. |
22 |
|
23 |
Remember the LSB half a year or so ago? Their main accomplishment was the FHS |
24 |
(in itself a very good thing). How did they characterize themselves back |
25 |
then? (Maybe they still do.) They would provide standard specifications for |
26 |
various parts of Linux, as they did with the FHS, so that the distributions |
27 |
become interoperable! I always thought this meant I could compile - well, |
28 |
anything - even a program designed for some other Unix perhaps - and it would |
29 |
work out of the box! But is seems that the LSB, after debating for what - a |
30 |
year, more? - finally decided that the best way to ensure that would be to |
31 |
make other people compile for me, and give me binary RPMs! If it took a year |
32 |
to decide, it must have been 'stuck in committee'... |
33 |
|
34 |
The distributions are supposed to be different - that's why they exist. |
35 |
Grassroots anti-LSB movement anyone?.. :-) |
36 |
|
37 |
Dan Armak |
38 |
|
39 |
On Monday 09 July 2001 22:24, you wrote: |
40 |
> Dan Armak <ermak@×××××××××××××.il> writes: |
41 |
> > But there's already one such method that always works - configure; |
42 |
> > make; make install. If LSB says RPMs are better than that, it |
43 |
> > discourages practicing what is the heart of Portage - automatized |
44 |
> > downloading, compiling & installing. The LSB should push for |
45 |
> > standardized results, not for a standard way of achieving them. |
46 |
> |
47 |
> extremely well put. |
48 |
> |
49 |
> > Whoever wants a pre-compiled package will eventually be able to get |
50 |
> > it via Portage which already supports binary packages. Whoever gets |
51 |
> > a package from its home site as source is thus encouraged to write |
52 |
> > an ebuild for it and give back to the community. RPM availability |
53 |
> > would desatroy that - Portage and emerge would simply become much |
54 |
> > less important. |
55 |
> |
56 |
> well, and there is more icky stuff: |
57 |
> |
58 |
> ,----[ <url: http://www.linuxbase.org/spec/gLSB/gLSB/x12251.html > ] |
59 |
> |
60 |
> | Package Dependencies |
61 |
> | |
62 |
> | Packages must depend on a dependency "lsb". They may not depend on |
63 |
> | other system-provided dependencies. If a package includes "Provides" |
64 |
> | it must only provide a virtual package name which is registered to |
65 |
> | that application. |
66 |
> |
67 |
> `---- |
68 |
> |
69 |
> at first, one might think, "great, no more looking for the oddball |
70 |
> package that contains <foo>"... but in reality, you're saying. |
71 |
> "bundle everything inside lsb and everything outside as well". since |
72 |
> there isn't a _real_ frontend like portage or apt for standard rpm |
73 |
> usage these days, every distro will need to make all their base |
74 |
> packages lsb-noted, but who'll _do_ that? and what will lsb do when |
75 |
> debian, slackware and Suse come along saying "hey, we want _this_ to |
76 |
> be the glibc-package", but RedHat already has a "lsb-glibc"-package? |
77 |
> |
78 |
> you don't want _your_ lsb-packages to depend on other distros |
79 |
> lsb-packages do you? |
80 |
> |
81 |
> > Of course, choice is important. So whoever thinks RPMs are good for |
82 |
> > Gentoo can go ahead and modify Portage/emerge to support them. |
83 |
> |
84 |
> agreed. being able to say "emerge -rpm <package>" might not be a bad |
85 |
> thing, but it's still not as nice. the only _real_ reason for a |
86 |
> common binary format is for the business world who want to be able |
87 |
> to brand their binary package as "lsb-approved". of course, the way |
88 |
> they'll do this is called "static linking", just to be on the safe |
89 |
> side. I doubt we'll see this change. |
90 |
> |
91 |
> > But people who still think actually compiling a package with the |
92 |
> > correct optimizations for you CPU is best <gasp> shouldn't be |
93 |
> > branded non-standard. (Or non-mainstream <gasp>). |
94 |
> |
95 |
> it's been that way for a while. personally I've used redhat, some |
96 |
> debian, some suse and some slackware for some time. I like different |
97 |
> things from different places, and I to love the _idea_ behind |
98 |
> Gentoo, because it addresses everything I've missed. easy to |
99 |
> customize, easy to upgrade, easy to admin and still state of the art |
100 |
> where you want it to be so (sadly, debian doesn't make the last |
101 |
> point at all). |
102 |
> |
103 |
> > Well, that's my opinion, for what it's worth. (phew!) |
104 |
> |
105 |
> right, that means we're up to what? 0.04$? :) |